Quantcast
Channel: Hacker News
Viewing all 25817 articles
Browse latest View live

Bitcoin’s Price Was Artificially Inflated Last Year, Researchers Say

$
0
0

SAN FRANCISCO — A concentrated campaign of price manipulation may have accounted for at least half of the increase in the price of Bitcoin and other big cryptocurrencies last year, according to a paper released on Wednesday by an academic with a history of spotting fraud in financial markets.

The paper by John Griffin, a finance professor at the University of Texas, and Amin Shams, a graduate student, is likely to stoke a debate about how much of Bitcoin’s skyrocketing gain last year was caused by the covert actions of a few big players, rather than real demand from investors.

Many industry players expressed concern at the time that the prices were being pushed up at least partly by activity at Bitfinex, one of the largest and least regulated exchanges in the industry. The exchange, which is registered in the Caribbean with offices in Asia, was subpoenaed by American regulators shortly after articles about the concerns appeared in The New York Times and other publications.

Mr. Griffin looked at the flow of digital tokens going in and out of Bitfinex and identified several distinct patterns that suggest that someone or some people at the exchange successfully worked to push up prices when they sagged at other exchanges. To do that, the person or people used a secondary virtual currency, known as Tether, which was created and sold by the owners of Bitfinex, to buy up those other cryptocurrencies.

“There were obviously tremendous price increases last year, and this paper indicates that manipulation played a large part in those price increases,” Mr. Griffin said.

Bitfinex executives have denied in the past that the exchange was involved in any manipulation. The company said on Wednesday that it had never engaged in “any sort” of market or price manipulation. “Tether issuances cannot be used to prop up the price of Bitcoin or any other coin/token on Bitfinex,” Jan Ludovicus van der Velde, Bitfinex’s chief executive, said in a statement.

The authors of the new 66-page paper do not have emails or documents that prove that Bitfinex knew about or was responsible for price manipulation. The researchers relied on the millions of transaction records that are captured on the public ledgers of all virtual currency transactions, known as the blockchain, to spot patterns. This method is not conclusive, but it has helped government authorities and academics spot suspicious activity in the past.

In particular, Mr. Griffin and Mr. Shams examined the flow of Tether, a token that is supposed to be tied to the value of the dollar and that is issued exclusively by Bitfinex in large batches. They found that half of the increase in Bitcoin’s price in 2017 could be traced to the hours immediately after Tether flowed to a handful of other exchanges, generally when the price was declining.

Other large virtual currencies that can be purchased with Tether, such as Ether and Zcash, rose even more quickly than Bitcoin in those periods. The prices rose much more quickly on exchanges that accepted Tether than they did on those that did not, and the pattern ceased when Bitfinex stopped issuing new Tether this year, the authors found.

Sarah Meiklejohn, a professor at the University College London who pioneered this sort of pattern spotting, said the analysis in the new paper “seems sound” after reviewing it this week.

Philip Gradwell, the chief economist at Chainalysis, a firm that analyses blockchain data, also said the study “seems credible.” He cautioned that a full understanding of the patterns would require more analysis.

Mr. Griffin previously wrote research pointing to fraudulent behavior in several other financial markets. He drew attention for a 2016 paper that suggested that a popular financial contract tied to the volatility in financial markets, known as the VIX, was being manipulated. A whistle-blower later came forward to confirm those suspicions, and now several active lawsuits are focused on the allegations.

Beyond his work at the University of Texas, Mr. Griffin has a consulting firm that works on financial fraud cases, including some in the virtual currency industry.

“The relationship between Tether and the price of Bitcoin has been flagged for months within the community,” said Christian Catalini, a professor at the Massachusetts Institute of Technology who specializes in blockchain research. “It is great to see academic work trying to causally assess if market manipulation is taking place.”

The new paper is not the first academic work to identify manipulation in the virtual currency markets. A paper published last year by a team of Israeli and American researchers said much of Bitcoin’s big price increase in 2013 was caused by a campaign of price manipulation at what was then the biggest exchange, Mt. Gox.


Dragonfly: Alibaba P2P file distribution system

$
0
0

README.md

Join the chat at https://gitter.im/alibaba/DragonflyLicenseFOSSA StatusGoDocGo Report CardBuild StatusCircleCIcodecov

Dragonfly

Contents

Introduction

Dragonfly is an intelligent P2P based file distribution system. It resolves issues like low-efficiency, low-success rate and waste of network bandwidth in large-scale file distribution scenarios such as application deployment, large-scale cache file distribution, data file distribution, image distribution etc. At Alibaba, the system transfers 2 billion times and distributes 3.4PB of data every month, it has become one of the most important piece of infrastructure at Alibaba. The reliability is up to 99.9999%.

DevOps takes a lot of benefits from container technologies, but at the same time, it also brings a lot of challenges: the efficiency of image distribution, especially when you have a lot of applications and require image distribution at the same time. Dragonfly works extremely well with both Docker and Pouch, and actually we are compatible with any other container technologies without any modifications of container engine.

It delivers up to 57 times the throughput of native docker and saves up to 99.5% the out bandwidth of registry.

Dragonfly makes it simple and cost-effective to set up, operate, and scale any kind of files/images/data distribution.

Features

The project is an open source version of the dragonfly and more internal features will be gradually opened.

  • P2P based file distribution: Using P2P technology for file transmission, which can make full use of the bandwidth resources of each peer to improve download efficiency, saves a lot of cross-IDC bandwidth, especially costly cross-board bandwidth
  • Non-invasive support all kinds of container technologies: Dragonfly can seamlessly support various containers for distributing images.
  • Host level speed limit: Many downloading tools(wget/curl) only have rate limit for the current download task,but dragonfly also provides rate limit for the entire host.
  • Passive CDN: The CDN mechanism can avoid repetitive remote downloads.
  • Strong consistency: Dragonfly can guarantee that all downloaded files must be consistent even if users do not provide any check code(MD5).
  • Disk protection and high efficient IO: Precheck Disk space, delay synchronization, write file-block in the best order, split net-read / disk-write, and so on.
  • High performance: Cluster Manager is completely closed-loop, which means, it does not rely on any DB or distributed cache, processing requests with extremely high performance.
  • Exception auto isolation: Dragonfly will automatically isolate exception nodes(peer or Cluster Manager) to improve download stability.
  • No pressure on file source: Generally, as long as a few Cluster Managers download file from the source.
  • Support standard http header: Support http header, Submit authentication information through http header.
  • Effective concurrency control of Registry Auth: Reduce the pressure of the Registry Auth Service.
  • Simple and easy to use: Very few configurations are needed.

Comparison

Test Environment
Dragonfly server2 * (24core 64GB 2000Mb/s)
File Source server2 * (24core 64GB 2000Mb/s)
Client4core 8GB 200Mb/s
Target file size200MB
Executed Date2016-04-20

For Dragonfly, no matter how many clients issue the file downloading, the average downloading time is always around 12 seconds. And for wget, the downloading time keeps increasing when you have more clients, and as the amount of wget clients reaches 1200, the file source will crash, then it can not serve any client.

License

Dragonfly is available under the Apache 2.0 License.

Commercial Support

If you need commercial support of Dragonfly, please contact us for more information: 云效.

Dragonfly is already integrated with AliCloud Container Services If you need commercial support of AliCloud Container Service, please contact us for more information: Container Service

Intel FP security issue

$
0
0

Delft scientists make first “on demand” entanglement link

$
0
0

News - 13 June 2018 - Communication

Researchers at QuTech in Delft have succeeded in generating quantum entanglement between two quantum chips faster than the entanglement is lost. Entanglement - once referred to by Einstein as "spooky action" - forms the link that will provide a future quantum internet its power and fundamental security. Via a novel smart entanglement protocol and careful protection of the entanglement, the scientists led by Prof. Ronald Hanson are the first in the world to deliver such a quantum link ‘on demand’. This opens the door to connect multiple quantum nodes and create the very first quantum network in the world. They publish their results on 14 June in Nature.

By exploiting the power of quantum entanglement it is theoretically possible to build a quantum internet that cannot be eavesdropped on. However, the realization of such a quantum network is a real challenge: you have to be able to create entanglement reliably, 'on demand', and maintain it long enough to pass the entangled information to the next node. So far, this has been beyond the capabilities of quantum experiments.

Scientists at QuTech in Delft have now been the first to experimentally generate entanglement over a distance of two metres in a fraction of a second, 'on demand', and subsequently maintain this entanglement long enough to enable -in theory- further entanglement to a third node. ‘The challenge is now to be the first to create a network of multiple entangled nodes: the first version of a quantum internet’, professor Hanson states. 

In 2015, Ronald Hanson's research group already became world news: they were the first to generate long-lived quantum entanglement over a distance (1.3 kilometres), , allowing them to providefull experimental proof of quantum entanglement for the first time. This experiment is the basis of their current approach to developing a quantum internet: distant single electrons on diamond chips are entangled using photons as mediators.

However, so far this experiment has not had the necessary performance  to create a real quantum network. Hanson: 'In 2015 we managed to establish a connection once an hour, while the connection only remained active for a fraction of a second. It was impossible to add a third node, let alone multiple nodes, to the network.’

The scientists have now made multiple innovative improvements to  the experiment. First of all, they demonstrated a new entanglement method. This allows for the generation of entanglement forty times a second between electrons at a distance of two metres. Peter Humphreys, an author of the paper, emphasises: 'This is a thousand times faster than with the old method.’ In combination with a smart way of protecting the quantum link from external noise, the experiment has now surpassed a crucial threshold: for the first time, entanglement can be created  faster than it is lost.

Through technical improvements, the experimental setup is now always ready for 'entanglement-on-demand'. Hanson: 'Just like in the current internet, we always want to be online, the system has to entangle on each request.’ The scientists have achieved this by adding smart quality checks. Humphreys: 'These checks only take a fraction of the total experimental time, while allowing us to ensure that our system is ready for entanglement, without any manual action'.

The researchers already demonstrated last year that they were able to protect https://qutech.nl/one-step-closer-to-the-quantum-internet-by-distillation/a quantum entangled link while a new connection was generated.  By combining this and their new results, they are ready to create quantum networks with more than two nodes.  The Delft scientists now plan to realize such a network between several quantum nodes. Hanson: 'In 2020, we want to connect four cities in the Netherlands via quantum entanglement. This will be the very first quantum internet in the world.’

This work was supported by the Netherlands Organisation for Scientific Research (NWO) through a VICI grant and by the European Research Council through a Starting Grant and a Synergy Grant.

Through-Wall Human Pose Estimation Using Radio Signals

$
0
0
Through-Wall Human Pose Estimation Using Radio Signals

Overview:


RF-Pose provides accurate human pose estimation through walls and occlusions. It leverages the fact that wireless signals in the WiFi frequencies traverse walls and reflect off the human body. It uses a deep neural network approach that parses such radio signals to estimate 2D poses. RF-Pose is trained using state-of-the-art vision model to provide cross-modal supervision. Once trained, RF-Pose uses only the wireless signal for pose estimation. Experimental results show that, when tested on visible scenes, the radio-based system is almost as accurate as the vision-based system used to train it. Yet, unlike vision-based pose estimation, the radio-based system can estimate 2D poses through walls despite never trained on such scenarios.

© Massachusetts Institute of Technology

Comcast, in 2nd Try, Offers $65B Cash for 21st Century Fox

$
0
0

Comcast announced an offer worth $65 billion for the bulk of 21st Century Fox’s businesses on Wednesday, setting up a showdown with the Walt Disney Company for Rupert Murdoch’s media empire.

The all-cash bid by Comcast, the largest cable company in the United States, came a day after a federal judge approved a merger between AT&T and Time Warner. Comcast executives had awaited the decision in that case before mounting their bid for 21st Century Fox.

In December, Disney struck a $52.4 billion, all-stock deal for Fox’s assets. Comcast, whose roughly $60 billion offer for the Fox assets was rebuffed last year, is now including contractual assurances such as a reverse breakup fee — worth about $2.5 billion — in the event a transaction is blocked by the government.

Comcast’s new offer is about 19 percent higher than Disney’s proposal, according to its statement.

Mr. Murdoch and his company’s board had rejected Comcast’s earlier offer partly on concerns the government would block the deal. But the AT&T-Time Warner decision also allayed many concerns that a Comcast takeover of 21st Century Fox’s businesses would be denied by regulators.

Brian L. Roberts, the head of Comcast, needed to move quickly. Fox shareholders are scheduled to vote on the Disney deal on July 10, but that date will be moved back if Mr. Murdoch and the Fox board decide to support Comcast’s offer. Disney would then have five days to respond with a counter bid.

The businesses that Mr. Murdoch has agreed to sell include the 20th Century Fox film and TV studios, almost two dozen regional sports networks like the Yankees’ YES channel, a lineup of cable networks that includes FX and a 30 percent ownership stake in the streaming service Hulu.

But the key attractions for Comcast are Fox’s broad international assets, which include its 39 percent stake in the European pay TV operator Sky and its control of Star, one of India’s largest media companies, which reaches 700 million people every month, according to the company.

Mr. Murdoch’s overseas business accounts for 27 percent of annual sales, about $7.8 billion. Comcast, whose cable business is strictly a domestic operation, draws in only 9 percent of its revenue from foreign agreements, largely through its NBCUniversal division.

Comcast has already made an offer to buy the other 61 percent of Sky in a separate deal. The Fox News cable network, the Fox broadcast stations, the Fox Business Network and the sports network FS1 would not be part of a transaction.

“This has all the makings of a very aggressive bidding war,” said Craig Moffett, co-founder of research firm MoffettNathanson and a longtime media analyst.

There is bad blood between Disney and Comcast. The rancor stretches back to at least 2004, when Comcast tried to swallow Disney. The Disney board fought off that attempt, but the chief executive Robert A. Iger and his top lieutenants have never forgotten it. Anyone riding the Jurassic Park rides at NBCUniversal’s theme parks can see what Comcast thinks of Mickey Mouse; one of Disney’s famous mouse ear hats floats in the water next to a raft that has been destroyed by a marauding dinosaur.

That failed deal looms over the current fight for Fox.

“Comcast seems hellbent on winning this time, and I think the narrative in Philadelphia is that Brian should have listened to his gut in 2004 and bought Disney,” Mr. Moffett said, referring to Comcast’s headquarters. “He seems very personally committed to this.”

Translating SQL Queries using map, reduce, and filter in JavaScript

$
0
0

Exercise: Suppose the following tables are given in the form of arrays of objects:

Translate the following SQL query using map, reduce, and filter:

You are not allowed to use loops, if statements, logical operators, or the ternary operator.

Solution:

We need to list all character names and the sum of the value of their items.

As characters and inventory are arrays, we can use the map, reduce, and filter.

The projection

corresponds to a simple map operation:

The next step is to join the characters and the inventory tables by filtering items belonging to a given character. We can also return the number of items belonging to a character in this step.

We have to return the name just like before, but we also have to find the corresponding items.

If you execute this code in the console, you can see that all three characters are there with their correct item count:

We only have one step left: summing the value of the items.

The SUM function is implemented by a reduce call in JavaScript:

The result is:

During the task, it is worth looking up the signature of reduce if you forgot it. Alternatively, you can reverse engineer it using a simple example if you get stuck and your interviewers insist in you not opening other tabs.

Sometimes you might have to use a testing platform to submit an answer, and the software might monitor when you leave a tab.

This is a stupid way of evaluating candidates, because it puts you in a big advantage if you have two computers side-by-side. You can execute any google searches you want in one computer, and solve the task in another. This is one reason why I believe all tests should be open book tests. If you want to browse the Internet, go for it! You have limited time, so you won’t be able to learn the basics of JavaScript while solving a task anyway.

This exercise is solved. Notice the analogy between SQL SELECT statements and map, reduce, filter. These are common questions not only in JavaScript.

A newcomer’s (angry) guide to R

$
0
0

Tim Smith<arrgh@tim-smith.us>, @biotimylated
with Kevin Ushey <kevinushey@gmail.com>, @kevin_ushey

R is a shockingly dreadful language for an exceptionally useful data analysis environment. The more you learn about the R language, the worse it will feel. The development environment suffers from literally decades of accretion of stupid hacks from a community containing, to a first-order approximation, zero software engineers. R makes me want to kick things almost every time I use it.

But there are a lot of great tools that are built in R. ggplot2 is first-in-class and Bioconductor packages are often essential. Sometimes there’s aught to do but grin and bear (though never without a side of piss and moan).

The documentation is inanely bad. I can’t explain it. aRrgh is my attempt to explain the language to myself. aRrgh exists as a living document and will continue to grow – it is not complete, but it got to a point where it seemed like it was probably useful so I decided to toss it on the web. It should be correct and it’s a bug if it isn’t. Please email me or file issues on Github.

The goal of the document is to describe R’s data types and structures while offering enough help with the syntax to get a programmer coming from another, saner language into a more comfortable place.

  • Basic syntax: crash course and gotchas; finding help
  • Atomic vectors: R’s simplest data types; logic values; vectorizing; arrays
  • Factors: A useful and misunderstood data type. Where they come from, how to handle them
  • Data frames: R’s structure for tabular data. How to create them, access semantics

To come?:

  • Indexing
  • Lists
  • Namespaces
  • Beyond Base R, Or: How I Learned To Stop Worrying and Love the Hadleyverse

© Tim Smith 2012-6. This work is made available under a Creative Commons Attribution-ShareAlike 3.0 Unported License.

If you enjoyed this, you will probably enjoy PHP: A Fractal of Bad Design, which is even more cathartic.


Sexual harassment is rife in the sciences, finds landmark US study

$
0
0

Sexual harassment is pervasive throughout academic science in the United States, driving talented researchers out of the field and harming others’ careers, finds a report from the US National Academies of Sciences, Engineering, and Medicine in Washington DC. The analysis concludes that policies to fight the problem are ineffective because they are set up to protect institutions, not victims — and that universities, funding agencies, scientific societies and other organizations must take stronger action.

“The cumulative effect of sexual harassment is extremely damaging,” says Paula Johnson, president of Wellesley College in Massachusetts and co-chair of the committee that wrote the report. “It’s critical to move beyond the notion of legal compliance to really addressing culture.”

The report, released on 12 June, is the most comprehensive study yet on the extent of harassment in the sciences. It comes in the wake of the #MeToo movement against sexual assault and harassment, and as the US national academies are grappling with whether to impose sanctions against members accused of harassment.

Notably, the report finds that the main mechanism for reporting sexual harassment on US campuses — Title IX, the federal law enacted in 1972 that outlaws discrimination on the basis of gender — has not reduced the incidence of sexual harassment. Institutions can find ways to comply with Title IX that avoid liability but don’t actually prevent harassment, says Asmeret Asefaw Berhe, a biogeochemist at the University of California, Merced. She is part of a national team working to train bystanders to intervene when they witness harassment, aiming to better prevent it.

The report’s many recommendations include: that research institutions should act to reduce the power differential between students and faculty members, perhaps by introducing group-based advising; that the government should prohibit confidentiality in settlement agreements, so that harassers cannot switch jobs without their new employer knowing about past behaviour; and that research organizations should treat sexual harassment at least as seriously as research misconduct.

“This is an incredibly comprehensive and ambitious report,” says Anna Bull, a sociologist at the University of Portsmouth, UK, and co-founder of The 1752 Group, which works to end harassment in academia. “They get beyond the ‘one bad apple’ approach and look at the culture that enables that one bad apple.”

Tracking the toll

The most common type of sexual harassment is gender harassment, the report finds. Such behaviour conveys the impression that women do not belong in the workplace or do not merit respect — “the put-downs as opposed to the come-ons”, Johnson says. These actions might seem minor but can seriously affect the person targeted. She says that they also set the stage for the other types of sexual harassment: unwanted sexual attention and sexual coercion.

All three varieties of sexual harassment are illegal in the United States when they interfere with a person’s work environment, yet all are common in science, engineering and medicine. Previous research has shown that the prevalence of sexual harassment in US academia, at 58%, is second only to the military’s 69%, and outpaces that of industry and government1. Women of colour experience particularly high rates of harassment2, as do people from sexual- and gender-minority groups3,4.

Men in academia also experience sexual harassment, although at lower rates than women; one study published in 2016 found that female graduate students at a public university in the US Pacific Northwest were 1.64 times more likely than male graduate students to have been sexually harassed by faculty members or staff3.

To build on those earlier studies, the academies’ committee commissioned an analysis that found that 20% of female science students at the University of Texas’s campuses reported being sexually harassed by faculty members or staff there. A similar survey of the Pennsylvania State University system concluded that 43% of graduate students experienced harassment.

All types of harassment, including gender harassment, can prove corrosive to scientists’ career development, according to interviews of 40 women faculty members conducted for the new report. One woman who had been raped by a colleague gave up research; another, who had been verbally berated by her dean, felt the experience derailed her from ever becoming a full professor.

“It’s not okay to treat your co-workers like dirt,” says Kathryn Clancy, a biological anthropologist at the University of Illinois at Urbana-Champaign and a member of the report committee. But university leaders often minimized or ignored the harassing behaviour, survey participants reported, especially when it involved higher-ranking faculty members who were perceived as stars in their department.

Searching for solutions

An institution’s workplace climate is by far the greatest predictor of sexual harassment, the academies’ report says. Title IX and related laws are a good start, says Clancy, but universities need to embrace other methods of addressing sexual harassment. These include ways for victims to report incidents without being re-traumatized or subject to retaliation.

“Many targets of harassment are women and minorities in vulnerable positions,” says Akiko Iwasaki, an immunologist at Yale University in New Haven, Connecticut. “If they feel like their careers rely on future recommendation letters from the harassers, they are less likely to want to come forward.” One possible solution mentioned in the academies’ analysis is an online reporting tool called Callisto, now available at 13 US universities. It allows people to file complaints securely and confidentially, and gives users the option to escalate a report only if another person names the same perpetrator.

Clancy also points to Britain’s Athena SWAN (Scientific Women’s Academic Network) programme, which grants bronze, silver or gold awards to institutions and departments on the basis of how well they meet gender-equity and -diversity goals. At least one UK funding agency requires institutions that receive its grant money to obtain an Athena SWAN certification. In the United States, the American Association for the Advancement of Science aims to institute similar awards through a new programme called SEA Change.

Culture shift

However strong the recommendations in the academies’ report, it is still up to universities to interpret them, says Jessica Cantlon, a cognitive neuroscientist who is in the process of leaving the University of Rochester in New York. There, she was part of a group of faculty members who sued university leadership over its handling of sexual-harassment allegations against a researcher in her department; the case is ongoing. “We are still waiting for tangible changes at our university, despite having voiced similar recommendations over two years ago in the wake of multiple student complaints about sexual harassment by a faculty member,” she says.

The report comes as the national academies are facing controversy over their policies related to harassment. In early May, neuroscientist BethAnn McLaughlin of Vanderbilt University in Nashville, Tennessee, launched a petition requesting that the National Academy of Sciences expel people who have been sanctioned for sexual harassment, retaliation or assault. More than 3,500 people have signed it.

National Academy of Sciences president Marcia McNutt says the group’s governing council will consider proposed changes at its next meeting in August. “This is something we have to take seriously as an organization,” she says. But, she adds, the academy would probably not initiate its own investigation of a member — instead referring any complaints that it receives to the leadership of that person's university. “One is ongoing right now,” she says. “No, I won’t tell you who it is.”

Intel SA-00145: Lazy FP State Restore

$
0
0

Legal Notices and Disclaimers

Intel provides these materials as-is, with no express or implied warranties.

All products, dates, and figures specified are preliminary based on current expectations, and are subject to change without notice.

Intel, processors, chipsets, and desktop boards may contain design defects or errors known as errata, which may cause the product to deviate from published specifications. Current characterized errata are available on request.

Intel technologies’ features and benefits depend on system configuration and may require enabled hardware, software or service activation. Performance varies depending on system configuration. No computer system can be absolutely secure. Check with your system manufacturer or retailer or learn more at https://intel.com.

Some results have been estimated or simulated using internal Intel analysis or architecture simulation or modeling, and provided to you for informational purposes. Any differences in your system hardware, software or configuration may affect your actual performance.

Intel and the Intel logo are trademarks of Intel Corporation in the United States and other countries.

*Other names and brands may be claimed as the property of others.
Copyright © Intel Corporation 2018

From Zero to Core ML Model (hero) in Minutes

$
0
0

From Zero to Core ML Model in Minutes

Machine Learning can be difficult to get your head around as a programmer. But aside from all the advanced mathematics and tooling, perhaps most difficult of all is learning how to let go.

Many of us have spent years honing our craft of writing code: expressive, type-safe, unit-tested, refactored, clean, DRY code. So it’s tough to hear that pretty much anyone with enough data (and patience) can use machine learning to solve hard problems without understanding why or how it works.

But this isn’t news to you. In fact, ML is at the top of your list of things to learn next! You have Andrew Ng’s course bookmarked in Safari and a load of unread PDFs littering your Downloads folder. All you need is a free weekend and…

Well, if that strikes a bit close to home for you then you’ll love CreateML.

Create ML is a new framework that makes it easy to train machine learning models. How easy? Drag pictures of dogs into a “Dogs” folder and pictures of cats into a “Cats” folder, write a few lines of Swift code, wait a couple minutes to train the model and boom: you have a image classifier that can tell you whether a picture contains a cat or a dog. You can even do this for text classification or regression for data tables.

If you haven’t already, go ahead and watch the Introducing Create ML session from this year’s WWDC. It’s like magic.

Pulling files from labeled directories makes for a nice demo, but what if your data set isn’t so nicely organized? This article shows how you can use Create ML to train a text classifier that predicts the programming language of unknown source code by manually creating the corpus from a heterogeneous data set.

You can find the complete training script and a demo playgroundhere.


Acquiring a Corpus of Data

First things first: we need some examples of source code.

If you’re anything like the author, you might have thought to use GitHub search to find code by language. And if so, you’d eventually realize that the both the GitHub API (both the REST and GraphQL versions) requires code search results to be scoped by user or project. At that point, you’d probably wonder why you decided you needed to make this yourself before finding something off-the-shelf, like this project by Source Foundry.

Our corpus includes labeled code samples in C, C++, Go, Java, JavaScript, Objective-C, PHP, Python, Ruby, Rust, and Swift. Each directory in the project root corresponds to a language and contains flattened checkouts of a handful of popular open source repositories for that language:

$ tree -L 2 code-corpora/swiftcode-corpora/swift
├── alamofire
│   ├── Alamofire.h
│   ├── Alamofire.swift
│   ├── AppDelegate.swift
│   ├── AuthenticationTests.swift
# ...

Notice that Objective-C header in the Swift project, though. We can’t rely entirely on the top-level directories as labels for their contents, because most projects include other auxillary scripts and source files (as well as README, LICENSE, and other repository miscellany).

Our training script uses the containing directory and file extension to determine the correct label for each file in our corpus. For example,.h files in the c directory are labeled as C,.h files in the cc directory are labeled as C++, and any file with the .go extension is labeled as Go:

switch(directory,fileExtension){case("c","h"),(_,"c"):label="C"case("cc","h"),(_,"cc"),(_,"cpp"):label="C++"case(_,"go"):label="Go"// ...default:// Unknown, skip}

Training the Model

To build our data table, we recursively enumerate the contents of the corpus directory and append the contents of each source file that we can identify:

varcorpus:[(text:String,label:String)]=[]letenumerator=FileManager.default.enumerator(at:corpusURL,includingPropertiesForKeys:[.isDirectoryKey])!forcaseletresourceasURLinenumerator{guard!resource.hasDirectoryPath,letlanguage=ProgrammingLanguage(for:resource,at:enumerator.level),lettext=try?String(contentsOf:resource)else{continue}corpus.append((text:text,label:language.rawValue))}let(texts,labels):([String],[String])=corpus.reduce(into:([],[])){(columns,row)columns.0.append(row.text)columns.1.append(row.label)}letdataTable=tryMLDataTable(dictionary:["text":texts,"label":labels])

Our original implementationappended MLDataTable objects, instead of initializing a single data table from an accumulated array. We found this to have nonlinear performance characteristics, which caused training to take closer to an hour instead of a few minutes.

With our data table in hand, we use the randomSplit(by:seed:) method to segment our training and testing data. The former is used immediately, passed into the MLTextClassifier initializer; the latter will be used next to evaluate the model.

let(trainingData,testingData)=dataTable.randomSplit(by:0.8,seed:0)letclassifier=tryMLTextClassifier(trainingData:trainingData,textColumn:"text",labelColumn:"label")

Creating an MLTextClassifier object takes a while, but you can track the progress by tailing STDOUT:

Automatically generating validation set from 5% of the data.
Tokenizing data and extracting features
10% complete
20% complete
30% complete
40% complete
50% complete
60% complete
70% complete
80% complete
90% complete
100% complete
Starting MaxEnt training with 8584 samples
Iteration 1 training accuracy 0.285182
Iteration 2 training accuracy 0.946295
Iteration 3 training accuracy 0.988001
Iteration 4 training accuracy 0.997554
Iteration 5 training accuracy 0.998602
Iteration 6 training accuracy 0.999185
Iteration 7 training accuracy 0.999651
Iteration 8 training accuracy 0.999767
Finished MaxEnt training in 7.12 seconds

The resulting model seems large for what it can do, weighing in at 3MB. However, it’s able to classify a file in ~20ms, which should be fast enough for most use cases.

Evaluating the Model

Let’s see how our classifier performs by calling the evaluation(on:) method and passing the testingData that we segmented before.

letevaluation=classifier.evaluation(on:testingData)print(evaluation)

Accuracy

At the top of our evaluation, we get a summary with the number of examples, the number of classes, and the accuracy:

Number of Examples1138
Number of Classes10
Accuracy99.56%

99.56% accuracy. That’s good, right? Let’s dig into the numbers to get a better understanding of how this behaves.


When you print(_:) an MLClassifierMetrics object, it shows a summary of the overall accuracy as well as a confusion matrix and a precision / recall table.

Confusion Matrix

A confusion matrix is a tool for visualizing the accuracy of predictions. Each column shows the predicted classes, and each row shows the actual class:

 CC++GoJavaJSObj-CPHPRubyRustSwift
C122000000000
C++07300020000
Go103330000000
Java000137000000
JS00005500000
Obj-C00000970000
PHP00000095000
Ruby000000013600
Rust00000000730
Swift00000000012

100% accuracy would have values along the diagonal line where the predicted and actual classes match, and zeroes everywhere else. However, our accuracy isn’t perfect, so we have a few stray figures. From the table, we can see that Go was mistaken for C once and C++ was incorrectly labeled as Objective-C twice.

Precision and Recall

Another way of analyzing our results is in terms of precision and recall.

ClassPrecision(%)Recall(%)
C99.19100.00
C++100.0097.33
Go100.0098.94
Java100.00100.00
JavaScript100.00100.00
Objective-C98.91100.00
PHP100.00100.00
Ruby100.00100.00
Rust100.00100.00
Swift100.00100.00

Precision measures the ability of the model to identify only the relevant classification within a data set. For example, our model had perfect precision for C++ because it never misidentified any source files as being C++, however it has imperfect precision for C because it incorrectly identified a Go file as being C.

Recall measures the ability of a model to identify all of the relevant classifications within a data set. For example, our model had perfect recall for C because it correctly identified all of the C source code in the training data, and imperfect recall for C++ because it missed two C++ files in the training data.

Writing the Model to Disk

So, we have our classifier, we’ve evaluated it and found it to be satisfactory. The only thing left to do is to is write it to disk:

letmodelPath=URL(fileURLWithPath:destinationPath)letmetadata=MLModelMetadata(author:"Mattt",shortDescription:"A model trained to classify programming languages",version:"1.0")tryclassifier.write(to:modelPath,metadata:metadata)

All told, training, evaluating, and writing the model took less than 5 minutes:

$time swift ./Trainer.swift281.84 real       275.51 user         5.60 sys

Testing Out the Model in a Playground

In order to use our model from a Playground, we need to compile it first. For an iOS or Mac app, Xcode would automatically generate a programmatic interface for us. However, in a Playground, we need to do this ourselves.

Call the coremlc tool using the xcrun command, specifying the compile action on the .mlmodel file and target the current directory for the output:

$ xcrun coremlc compile ProgrammingLanguageClassifier.mlmodel .

Take the resulting .mlmodelc bundle (it’ll look like a normal directory in Finder) and move it into the Resources folder of your playground. You can use this to initialize a Natural Language framework NLModel to classify text using the predictedLabel(for:) method:

leturl=Bundle.main.url(forResource:"ProgrammingLanguageClassifier",withExtension:"mlmodelc")!letmodel=try!NLModel(contentsOf:url)

Now you can call the predictedLabel(for:) method to predict the programming language of a string containing code:

letcode="""
struct Plane: Codable {
    var manufacturer: String
    var model: String
    var seats: Int
}
"""model.predictedLabel(for:code)// Swift

The sample code project for this post wraps this up with a fun drag-and-drop UI, so you can easily test out your model with whatever source files you have littering your Desktop.

Screenshot of Classifier Example

Conclusion

There’s no way this actually works… right? Source code isn’t like other kinds of text, and weighing keywords and punctuation equally with comments and variable names is obviously a flawed approach.

The way classifiers work, our model may well be fixating on irrelevant details like license comments in the file header. Heck, that 99% accuracy we saw could be more a reflection of file similarity within the same project than of the model’s actual predictive ability.

All of that said, it might just be good enough.

Consider this: in under an hour, we went from nothing to a working solution without any significant programming. That’s pretty incredible.

Create ML is a powerful way to prototype new features quickly. If a minimum-viable product is good enough, then your job is done. Or if you need to go even further, there are all kinds of optimizations to be had in terms of model size, accuracy, and precision by using something like TensorFlow or Turi Create.

Ask HN: How to find a mental health professional?

$
0
0
Ask HN: How to find a mental health professional?
204 points by throwaway-186136 hours ago | hide | past | web | favorite | 114 comments
We've been having a lot of great discussions here on HN lately about mental health, and the suggestion to seek professional help rings true.

But, depressed and burnt out and feeling isolated, the project of even finding a therapist in the first place can be overwhelming.

I live in a large city, where there are literally thousands of therapists, psychologists, psychiatrists, licensed social workers, life coaches, you name it. About half of them even accept my insurance.

How does one even begin to narrow the options down? After asking my GP and getting no suggestions, I'm at a complete loss.


Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Can this cooler save kids from dying?

$
0
0

The big chill

| June 13, 2018

Two of the things I love most about my job are getting to see amazing innovations and talk to remarkable people. During a recent trip to New York, I got to check both boxes. I met a woman named Papa Blandine Mbwey who is using a revolutionary new invention to help more kids get vaccinated.

Blandine has worked as a vaccinator in a remote part of the Democratic Republic of the Congo for over a decade. Most days, she travels on foot to villages all over her region so she can vaccinate kids who live too far from a health clinic to make the trip themselves.

Blandine’s job is complicated by a simple fact: vaccines must be kept between 2 and 8° C. If they get too warm, they spoil. If they get too cold, the water in them freezes, and they can stop working. Vaccines must stay within this temperature range through each step of what’s called the “cold chain.”

imageimage

By the time Blandine reaches the children, the vaccines she’s carrying have traveled nearly 5,100 miles. They could have spoiled at any point during that journey, but vaccines are particularly at risk during the last two stops.

First there’s the health clinics where vaccinators like Blandine usually pick up their supply of vaccines. Many of these clinics are in areas with frequent power outages or no electrical grid at all, which means the refrigerators can’t always keep the vaccines cold.

But even if the vaccines survive the clinic, they still need to make it to the children. Most vaccinators carry them in ice-lined coolers. If you’ve used a cooler to keep your drinks cold at a picnic, you know the big problem with ice: it starts melting as soon as you take it out of the freezer. This means that some of the kids never get vaccinated, because coolers can’t keep vaccines cold long enough to reach them.

Several years ago, I asked a group of inventors called Global Good that I support to take on the cold chain problem. They came up with two remarkable innovations that are changing the game for vaccinators like Blandine.

The first is the MetaFridge. Although it looks like a regular refrigerator, MetaFridge has a hidden superpower: it keeps vaccines cold without power for at least five days. The electrical components are designed to keep working through power surges and brown-outs. During extended outages, an easy-to-read screen tells you how much longer it can stay cool without power so health workers know when to run a generator or move vaccines elsewhere. And if the fridge stops working properly, it transmits data remotely to a service team so they can fix it before vaccines are at risk of spoiling.

The other innovation Global Good invented is the Indigo cooler, which is the device you see Blandine using in the video above. It keeps vaccines at the right temperature for at least five days with no ice, no batteries, and no power required during cooling.

It sounds counterintuitive, but the Indigo needs heat before you can use it. When exposed to a heat source, water inside its walls evaporates and moves into a separate compartment. It can then sit on a shelf for months after heating, ready for use.

When it’s finally time to head out to the children, you open a valve, and the water starts moving back where it started. Because the pressure inside the Indigo has been lowered to the point where water evaporates at 5° C, the water particles take heat with them (the way sweating lowers your body temperature) and cool the storage area down to the perfect temperature for vaccine storage.

Both inventions are already making an impact in the field. A Chinese manufacturer started selling the MetaFridge last year, and a new solar-powered version will hit the market soon. One of the biggest surprises so far is just how much we’ve learned from its remote data monitoring capabilities. We knew the electrical grids in sub-Saharan Africa were unreliable, but we now know exactly how much the power fluctuates. This information will be helpful moving forward for health providers and anyone designing a product meant to work in these areas.

The Indigo is in the field trial phase. It’s still early, but the data suggests that the Indigo is allowing vaccinators to reach four times as many places as they could with the old ice-based coolers. That’s a big deal, and I’m excited to learn more.

Keeping vaccines cold when you’re delivering them to the most remote places on earth is a tough problem—and these devices show how innovation can help solve tough problems. I hope MetaFridge and Indigo inspire other inventors to find creative solutions.

"+(mainIIG+1)+"  of  "+listOfObjects.length+"

"+capt01+" "+capt02+""); }else{ $('.InlineImageGalleryText').html("

"+(mainIIG+1)+"  of  "+listOfObjects.length+"

"+capt01+" "+capt02+""); } $('.iigNextArrow').click(function() { iignext(); }) $('.iigPrevArrow').click(function() { iigprev(); }) adjustIIG(); } function iigprev(){ var ww = $(window).width(); mainIIG-=1; if (mainIIGlistOfObjects.length-1){nIIG=0;} $('.InlineImageGalleryNextImage').html(listOfObjects[nIIG].outerHTML); pIIG = mainIIG-1; if (pIIG

"+(mainIIG+1)+"  of  "+listOfObjects.length+"

"+capt01+" "+capt02+""); }else{ $('.InlineImageGalleryText').html("

"+(mainIIG+1)+"  of  "+listOfObjects.length+"

"+capt01+" "+capt02+""); } $('.iigNextArrow').click(function() { iignext(); }) $('.iigPrevArrow').click(function() { iigprev(); }) adjustIIG(); } function adjustPQ(){ var pqW = $('.suspendedR').width(); var pqH = $('.suspendedR').height(); $('.pqrq').css("margin-left",pqW-60); $('.pqrq').css("margin-top",pqH-110); } function adjustIIG(){ var ww = $(window).width(); var iigW = $('.InlineImageGalleryBase').width(); //console.log(ww); //if (iigWlistOfObjects.length-1){nIIG=0;} $('.InlineImageGalleryNextImage').html(listOfObjects[nIIG].outerHTML); pIIG = mainIIG-1; if (pIIG

"+(mainIIG+1)+"  of  "+listOfObjects.length+"

"+capt01+" "+capt02+""); $('.IIGheadline').css("width",iigMW-200); var headH = $('.IIGheadline').height(); if (headH==15){ $('.IIGheadline').css("top",17); }else if(headH==30){ $('.IIGheadline').css("top",9); }else if(headH==45){ $('.IIGheadline').css("top",2); }else{ $('.IIGheadline').css("top",17); } $('.InlineImageGalleryText').css("height", 50); //console.log("h: "+$('.IIGheadline').height()+" w: "+$('.IIGheadline').width()); } } // ADD SUPPORT FOR INLINE INTERACTIVE FILES function formatInlineElems(){ // related content //small-12 medium-12 large-12 columns var listOfRelated = []; $('.promounit').each(function (i, obj) { listOfRelated.push(obj); }); //console.log(listOfRelated); // CTA $('.bgcta').each(function (i, obj) { //console.log($(this).children().eq(1).prop('src')); $(this).after( "" ); var imgOn = $(this).children().eq(1).prop('src'); var imgOff = $(this).children().eq(0).prop('src'); $("#bgctab"+i).mouseenter(function() { $(this).find('img').attr("src", ""+imgOn); }).mouseleave(function() { $(this).find('img').attr("src", ""+imgOff); }); $(this).remove(); $("#bgctab"+i).children(0).attr("width", "100%" ); }); // marginalia right $('.bgsbnoter').each(function (i, obj) { var ww = $(window).width(); if (ww > 980) { $(this).after( "" ); $(this).remove(); $("#bgsbnoter"+i).children(0).attr("width", "100%" ); $("#bgsbnoter"+i).css("position","relative"); $("#bgsbnoter"+i).css("top",$(this).attr("offsetY")); $("#bgsbnoter"+i).css("left",$(this).attr("offsetX")); } else { $(this).remove(); } }); // marginalia left $('.bgsbnotel').each(function (i, obj) { var ww = $(window).width(); if (ww > 980) { $(this).after( "" ); $(this).remove(); $("#bgsbnotel"+i).children(0).attr("width", "100%" ); $("#bgsbnotel"+i).css("position","relative"); $("#bgsbnotel"+i).css("top",$(this).attr("offsetY")); $("#bgsbnotel"+i).css("left",$(this).attr("offsetX")); } else { $(this).remove(); } }); // left image $('.bgli').each(function (i, obj) { //$(this).after( "

"+listOfRelated[0].outerHTML+"

" ); // related content $(this).after( "" ); $(this).remove(); $("#gblib"+i).children(0).attr("width", "100%" ); }); // block quote $('.bgbq').each(function (i, obj) { var PQtext = $(this).html(); var s = PQtext.replace(/\u201C/g, " "); var s2 = s.replace(/\u201D/g, ""); $(this).after( "" ); $(this).remove(); $(".articleContent p").css("z-index", "1"); $(".articleContent p").css("position", "relative"); }); // underline $('.TGNil_underline').each(function (i, obj) { $(this).css("border-bottom","0px solid "+"#f5d840");//+$(this).attr("mycolor")); $(this).css("box-shadow","inset 0 -5px 0 "+"#f5d840");//+$(this).attr("mycolor")); $(this).css("color","inherit"); $(this).css("-webkit-transition","background .15s cubic-bezier(.33,.66,.66,1)"); $(this).css("transition","background .15s cubic-bezier(.33,.66,.66,1)"); }); // highlight $('.TGNil_highlight').each(function (i, obj) { $(this).css("border-bottom","2px solid "+$(this).attr("mycolor")); $(this).css("box-shadow","inset 0 -23px 0 "+$(this).attr("mycolor")); $(this).css("color","inherit"); $(this).css("-webkit-transition","background .15s cubic-bezier(.33,.66,.66,1)"); $(this).css("transition","background .15s cubic-bezier(.33,.66,.66,1)"); }); // inline image gallery $('.bgiig').css("display", "block");//remove after testing $('.bgiig').each(function (i, obj) { $(this).children().each(function () { $(this).attr('style', 'height:100%!important;width:auto!important;'); listOfObjects.push(this); }); var capt01 = listOfObjects[0].getAttribute('data-caption1');//listOfObjects[0].dataset.caption1 var capt02 = listOfObjects[0].getAttribute('data-caption2'); if (capt01==null){capt01="";} if (capt02==null){capt02="";} //thumb.getAttribute('data-bigwidth') $(this).after( "

"+listOfObjects[0].outerHTML+"

1  of  "+listOfObjects.length+"

"+capt01+" "+capt02+"
"); //$(this).after( "

"+listOfObjects[0].outerHTML+"

1  of  "+listOfObjects.length+"

"+listOfObjects[0].dataset.caption1+" "+listOfObjects[mainIIG].dataset.caption2+"
" ); $(this).remove(); //console.log(listOfObjects); //$('.articleContent').css("color","#F00"); }); var IIGlength = listOfObjects.length; //console.log(IIGlength); $('.InlineImageGalleryNext').click(function() { iignext(); }) $('.InlineImageGalleryPrev').click(function() { iigprev(); }) $('.iigNextArrow').click(function() { iignext(); }) $('.iigPrevArrow').click(function() { iigprev(); }) // if( $('.InlineImageGalleryBase').length ) { adjustIIG(); } // if( $('.suspendedR').length ) { adjustPQ(); } //$('.suspended').each(function (i, obj) { // var thisW = $(this).width(); // var thisH = $(this).height(); // if(thisW>thisH){ // $(this).css("width",450); // } //}); } function getCommentCount() { var see = $(".disqusLink").text(); if (see == "") { setTimeout(getCommentCount, 500); } else { var i = parseInt(see); $(".commentCount").each(function (ii, v) { $(v).text(i.toString()); }); } } var scrollwidthDU = function () { var inner = document.createElement('p'); inner.style.width = "100%"; inner.style.height = "200px"; var outer = document.createElement('div'); outer.style.position = "absolute"; outer.style.top = "0px"; outer.style.left = "0px"; outer.style.visibility = "hidden"; outer.style.width = "200px"; outer.style.height = "150px"; outer.style.overflow = "hidden"; outer.appendChild(inner); document.body.appendChild(outer); var w1 = inner.offsetWidth; outer.style.overflow = 'scroll'; var w2 = inner.offsetWidth; if (w1 == w2) w2 = outer.clientWidth; document.body.removeChild(outer); return (w1 - w2); }(); function adjustPrevNext() { var hn = $(".nextText").height(); var hp = $(".prevText").height(); var h = Math.max(hn+24, hp+24); //console.log("prev " + hp + " next " + hn); $("#content_0_nextLink").css("height", h + "px"); $("#content_0_prevLink").css("height", h + "px"); //$(".nextText").css("padding-top", (((h-hn) / 2) + 8) + "px"); //$(".prevText").css("padding-top", (((h-hp) / 2) + 8) + "px"); $(".nextArrow").css("padding-top", (((h-33) / 2)) + "px"); $(".prevArrow").css("padding-top", (((h-33) / 2)) + "px"); } function doOmnitureEvent(event) { try { s.linkTrackVars = "events,eVar9"; s.linkTrackEvents = "event11,event12,event13,event15"; s.events = event; s.tl(true, 'o', 'article'); } catch (e) { //console.log(e.message); //console.log(e.name); } } $(document).ready(function () { // Omniture Stuff var tagsLocation = $(".bottom").offset().top; // was tagsLocation var scrolledToTags = false; var scrolledToComments = false; var tgnbody = $("#tgnbody"); $(".fbShareThis").click(function () { s.eVar9 = "facebook share"; doOmnitureEvent("event11"); }); $(".twitterShareThis").click(function () { s.eVar9 = "twitter share"; doOmnitureEvent("event12"); }); $("#tgnbody").scroll(function () { if (scrolledToTags === false) { if (tgnbody.height() + tgnbody.scrollTop() > tagsLocation) { scrolledToTags = true; doOmnitureEvent("event13"); } } if (scrolledToComments === false) { if (tgnbody.scrollTop() > tagsLocation) { scrolledToComments = true; doOmnitureEvent("event15"); } } }); }); $(document).ready(function () { var clicked = false; var title = document.location.pathname; if ($(".videoEmbed iframe").length > 0) { $(".videoEmbed iframe").iframeTracker({ blurCallback: function () { if (clicked === false) { try { s.events = "event8"; s.eVar6 = title; s.prop9 = title; s.linkTrackVars = 'events,eVar6,prop9'; s.linkTrackEvents = "event8"; s.tl(true, 'o', title); } catch (e) { //console.log(title); //console.log(event); //console.log(e.message); //console.log(e.name); } clicked = true; } return true; } }); } });

Librarian wins surprise judgement against Equifax in small claims court

$
0
0
Jessamyn West’s Twitter profile picture.

This article by Matt Hongoltz-Hetling was initially published by the Valley News.

RANDOLPH — In a small claims court ruling that surprised even the victor, a self-described member of the “librarian resistance” has won a $600 judgment against Equifax, the credit ratings agency that collects financial data on nearly a billion consumers and businesses worldwide.

In September, Jessamyn West, of Randolph, walked into the Orange County Courthouse in Chelsea and filed papers asking the judge to compel Equifax to pay her nearly $5,000 in connection with a data breach that affected more than 100 million people.

“I found out along with everyone else in September that Equifax had lost my information,” West said during a Tuesday phone interview. “They didn’t patch part of their computer system, and hackers absconded with the information of me and 140 million other Americans.”

The 49-year-old West said she’s not sure what of her personal information was compromised, though the company has admitted the hackers accessed names, addresses, Social Security numbers and, for hundreds of thousands of people, credit card information.

West is a tech-savvy librarian at the Randolph Technical Career Center who gives regular public lectures on privacy rights, online security and information science across the state.

“I’m a civics nerd. I’m a justice of the peace. It’s all very ‘Vermonty,’” she said.
In the aftermath of the breach, West was dismayed by what she perceived to be a general sense of defeatism when discussing the breach with friends.

“I would go to dinner and they would say ‘Equifax! Crazy! Nothing you can do,’” she recalled.

But for West, the situation rankled. Her mother had died that year and, as co-executor of the estate, West had run into various problems verifying her online identity during what already was a distressing experience.

So she decided to do something about it.

“Filing small claims cases is a thing human beings can do. It’s not that hard,” she said. “And the amount of money it cost, 90 bucks — it’s not nothing, but to me it’s not a huge amount of money.”

Jessamyn West
A tweet from Jessamyn West’s account announcing the news of her legal win against Equifax.

West didn’t expect Equifax to bother to put in an appearance during a May court date. The Atlanta-based company, one of the country’s “Big Three” credit ratings agencies, reported $3.4 billion in revenue for 2017, up 7 percent from 2016.

But someone did show up.

West described the company representative, who had been flown into the state, as a “surprisingly nice and friendly” paralegal, who, between the courtroom proceedings, chatted with her about novel-to-him Vermont experiences, including the state’s craft beers and dirt roads.

“I expected someone in a suit who did not have a sense of humor about having to come to Vermont,” she said. “Instead I got this guy.”

After West and the Equifax representative argued their respective positions, West said she thought she would lose the case because he had pointed out her difficulties in proving “speculative damages” — which cannot typically be recovered by a plaintiff.

Judge Bernard Lewis said he would look into the idea of speculative damages, and issue a ruling soon. On June 4, Lewis issued a court order that found West was owed money to cover the cost of up to two years of payments to online identity protection services, plus her $90 filing fee.

West said the victory shows that large corporations don’t always win.

“I’m not even the little guy,” she said. “I’m the microscopic, can’t-even-see-me, speck of dust guy.”

She encouraged others to follow her lead, “if they think they have a case. Especially if weird stuff is going on with your finances.”

In an email on Tuesday, Marisa Salcines, an Equifax spokeswoman, declined to comment on the case.

Though it’s unclear whether Equifax’s bottom line is being affected, news agencies across the country have reported a flurry of class action lawsuits and small claims court cases against the company in recent months. A Stanford University student made headlines nationwide when he created an online application that streamlined the process to file against the company, and some cases reportedly have been resolved against the company for up to $10,000 each.

West said that seeing news coverage about that application had helped spark her interest in filing the suit.

But now, West said, she’s moving on to express her ire in other ways, at other companies.
“Today,” she said, “I’m mad at Greyhound…”

Take the legal journey with West in this thread from her Twitter account:


Apple to Close iPhone Security Hole That Police Use to Crack Devices

$
0
0

The friction came into public view after the F.B.I. could not access the iPhone of a gunman who, along with his wife, killed 14 people in San Bernardino, Calif., in late 2015. A federal judge ordered Apple to figure out how to open the phone, prompting Timothy D. Cook, Apple’s chief executive, to respond with a blistering 1,100-word letter that said the company refused to compromise its users’ privacy. “The implications of the government’s demands are chilling,” he wrote.

The two sides fought in court for a month. Then the F.B.I. abruptly announced that it had found an undisclosed group to hack into the phone, for which it paid at least $1.3 million. An inspector general’s report this year suggested the F.B.I. should have exhausted more options before it took Apple to court.

Since then, two main companies have helped law enforcement hack into iPhones: Cellebrite, an Israeli forensics firm purchased by Japan’s Sun Corporation in 2006, and Grayshift, which was founded by a former Apple engineer in 2016. Law enforcement officials said they generally send iPhones to Cellebrite to unlock, with each phone costing several thousand dollars to open. In March, Grayshift began selling a $15,000 GrayKey device that the police can use to unlock iPhones themselves.

Apple has closed loopholes in the past. For years, the police used software to break into phones by simply trying every possible passcode. Apple blocked that technique by disabling iPhones after a certain number of wrong passcodes, but the Grayshift and Cellebrite software appear to be able to disable that Apple technology, allowing their devices to test thousands of passcodes, Mr. Green said.

Cellebrite declined to comment. Grayshift did not respond to requests for comment.

Opening locked iPhones through these methods has become more common, law enforcement officials said. Federal authorities, as well as large state and local police departments, typically have access to the tools, while smaller local agencies enlist the state or federal authorities to help on high-profile cases, they said.

Law enforcement agencies that have purchased a GrayKey device include the Drug Enforcement Administration, which bought an advanced model this year for $30,000, according to public records. Maryland’s state police have one, as do police departments in Portland, Ore., and Rochester, Minn., according to records.


“SKAM,” the Radical Teen Drama That Unfolds One Post at a Time

$
0
0

The first installment of the teen drama SKAM Austin” popped up on Facebook almost without warning, on April 24th, at 3:40 P.M. Central Standard Time. No advertising preceded it. No interviews with the actors or the director accompanied its début, and the clip had no production credits. It was as if the footage were just another update in your Facebook feed. The show—an American version of the Norwegian phenomenon “SKAM,” whose title means “shame”—did not take the form of conventional episodes. Viewers were instead offered an array of scenes, of varying lengths, shot in and around a high school in Texas’s capital. One clip was two minutes long; another was eight. These fragments began sporadically “dropping” on Facebook Watch, the social network’s entertainment portal, in accordance with the action of the show. If a couple got into a fight in school at 12:40 P.M. on a Monday, the clip showed up on the platform at exactly that time, creating the uncanny impression that you were watching something that was actually happening. If the producers posted a clip showing a student getting dressed for a party on a Saturday night, many young viewers would be doing the same thing.

The substance of the show wasn’t that different from “Riverdale”: it offered the usual roundelay of broken hearts, bruised feelings, and hookups. Teens kissed. They zoned out in class. They shared earbuds. But “SKAM Austin” had many hidden layers, and the producers wanted viewers to uncover them all. The characters, some of them played by local teen-agers, all had Instagram accounts, and, like real people’s, the posts offered insights into the characters’ pasts and their hopes for the future. Collectively, the video clips, photographs, and comments imbued the characters with a depth that not even flashbacks provide in conventional TV.

Soon after the first, six-minute clip of “SKAM” appeared on Facebook Watch, I developed a theory about several of the characters: long before April 24th, it seemed, Megan, a member of the school’s dance troupe, had stolen a boy named Marlon from her friend Abby, another dancer; Abby, in revenge, had shut Megan out of her life, and as a result Megan had quit the troupe. The only hint that the clip itself had offered about the girls’ relationship was a moment of Megan’s gaze lingering on Abby as she swept by with the other dancers. To decode the implications of this split-second image, I needed to do what we often do these days after meeting interesting strangers at a party: I scoured the characters’ social-media accounts. “SKAM” is a kind of detective show, rewarding the viewer who is a skilled online stalker.

Scrutiny of Abby’s Instagram posts suggested that she had scrubbed her account of traces of her friendship with Megan. But, as often happens with actual teen-agers, she had been inexpert in rewriting her history, forgetting to delete a video. It showed the two girls happily taking on the “mannequin challenge”—recording themselves suddenly freezing up and holding a tricky pose. Culturally attuned viewers would recall that such videos became a viral sensation at the end of 2016. This meant that the rupture had occurred sometime after that date.

As with all Internet products, once you establish a connection to “SKAM” it’s very hard to sever it. Facebook and Instagram send viewers constant reminders to log back in and stay up to date: “abby_taffy just posted a photo”; “SKAM Austin posted a new episode on Facebook Watch.” (These messages appeared on my phone’s lock screen next to announcements of my daughter’s Instagram posts about our family’s puppy.) The notices help viewers keep abreast of the basic story, but to get maximum pleasure from “SKAM” you must constantly burrow into the latest Instagram Stories or screenshots of texts. Internet viewing is always as much about what everyone else is watching and thinking as about what you’re watching and thinking—scholars talk about the medium’s “emotional contagion.” And “SKAM” is addictive in precisely the same way that social media can be addictive. If you miss out on too many details, you’ll feel as if you’d been demoted to sitting alone in the school cafeteria.

The fictional social media of “SKAM Austin” soon generated real social media—fervid discussion on everything from Tumblr to Twitter. For an obsessed viewer, there’s no limit to the amount of time that can be spent on “SKAM Austin” fan pages. The Internet, by leaving you feeling uniquely alone, paradoxically encourages human interaction. Megan and Marlon immediately became the cynosure of legions of online commenters, many of whom assessed the couple as if they were real. One poster wrote, “Not to get too deep and personal here, but I had an exchange with a friend who also happens to be an ex, and it made me think of Marlon and Meg, and I hadn’t realized it until today. It might be why I have such red flags about them.” She asked if anyone else felt the same way. Soon afterward, another poster wrote, “Relaaaaaaaaaaate.”

Conventional TV is a one-way street: you sit in front of a screen and watch an episode. Just as you must be static in order to finish watching it, the program itself is static: it had to be written, filmed, and edited to a conventional length. It represents a producer’s best guess about what will interest you (and, when there are commercials, an advertiser’s best guess about what viewers like you will buy). The model proved stable for more than fifty years, but it has crumbled in the age of YouTube, Facebook, and Twitter. According to a recent Nielsen report, millennials spend twenty-seven per cent less time watching TV programs (including streaming ones) than do older viewers. Every day, YouTube has an average of five billion views, more than a billion of them from mobile devices. The average teen-ager spends almost nine hours a day consuming media online, and sends or receives more than a hundred text messages.

There is a clear creative opportunity in this shift away from the network model. What if all these seemingly disparate activities and digital platforms could be marshalled into a single narrative—a Gesamtkunstwerk for the Internet age? Would it make the old-fashioned television episode seem as antique as black-and-white TV did once color sets appeared? The time seems right for an experiment like “SKAM.” In an era of short attention spans, it can seem atavistic to watch a half-hour series, let alone binge-watch it. “Engagement” is the key metric for the online industry—advertisers want to pay for how often you like, post, and click, rather than for how long you passively watch—and “SKAM,” with its cliffhangers and its multiple entry points, is designed to inspire passionate engagement. Fidji Simo, the head of Facebook Watch, told me, “ ‘SKAM’ was just the perfect fit for the kind of content we wanted to do more of.” Indeed, Facebook, which has been losing young users to YouTube and Snapchat, needs such programming to attract them. And what better way to advertise Facebook than by creating a show in which all the characters use Facebook?

Two weeks into the series, a five-minute clip introduced the heartthrob of the football team. The first GIF of him appeared online before the clip finished. By the end of the day, one poster had put up twenty-two images extracted from the footage. I happened to be watching the clip drop that day with the Facebook Watch social-media staff, and even they seemed surprised by the barrage of fan activity it sparked.

Depending on your point of view, “SKAM” is either ingenious or cynical in the ways it rewards audience engagement: if you follow the Instagram accounts of the characters, they will sometimes follow you back. With “SKAM,” you’re not only an integral part of the spectacle; you’re also a producer. The show’s creators monitor fan commentary and sometimes respond to it by changing plot details on the fly. Viewers, teased by Facebook and the creators into believing that they are being heard, and that what they’re seeing is true—or close enough—experience “SKAM” less as an alternative reality than as an extension of their own lives. By inserting a story so skillfully into our digital domains, and keeping us endlessly tethered to that story, “SKAM” may be the future of TV.

In April, a few days before the first clip was posted, I met Julie Andem, the Norwegian creator of “SKAM,” at a café in Austin, across the street from her production office. Facebook Watch had relocated her from Oslo to oversee the American production, and she was still writing and directing new scenes. A key reason that “SKAM” felt fresh to viewers, she told me, was that its clips are shot very shortly before they air. Among other things, this approach allowed Andem to take into account fan feedback and contemporary events. After the show started, a character posted that he wanted tickets for an upcoming Kendrick Lamar concert in Austin. Lamar performed in the city on May 18th, and the story that day revolved around the concert. To heighten the sense that “SKAM” is unspooling in real time, Andem adopts visual techniques that mimic the latest fads on social media. One of the characters posted a makeup tutorial that looks just like those currently popular on YouTube. The timeliness of the characters’ status updates can be unsettling. On the day of the recentschoolshooting in Santa Fe, Texas, two characters put up distraught Instagram posts. One of them posted a map of Texas with a heart over Santa Fe and the caption “Why tf does this keep happening?” (Audience members responded with emotion: “Because we live in a country that thinks owning a gun is far more important than the lives of innocent kids”; “We need to stand together and fight until real change happens!”)

Andem had come to the café with her social-media director, Mari Magnus, a fellow-Norwegian, who plays a crucial role in the show’s production. Andem shoots the main show; Magnus shoots the Instagram Stories—collages of video, text, and photographs—that play off Andem’s scripts. Magnus also posts comments on Instagram in the guises of the characters. (“It is a bit creepy,” Andem has said of this ventriloquism.) Dressed in black tops and pants, the two women looked like European tourists on their way to Marfa, but they had been in Austin since October, doing very little besides working on the show.

Simon Fuller, the creator of “American Idol,” who bought the English-language rights to “SKAM” in 2016 and then partnered with Facebook Watch, told me that he had made the deal because he was impressed by Andem’s sensibility. “To be honest with you, I couldn’t see past Julie,” he told me. Andem, who until now has worked only in Norway, seems to have little interest in Hollywood fame. The executive producer of the original “SKAM” told me that Andem was unusually gifted at directing young people. When I asked Andem about this, she said, “Yes, I probably have an instinct, but I’m not aware of what I do.” She noted, “As soon as you start to comment on your own work, then some of the magic of the story goes away. Audiences want their own experience.” Her deflections were consistent with how “SKAM” feels: a viewer experiences the show less as the vision of a single auteur than as a vision intended for a single viewer.

I had expected Andem to tell me about her struggle to create the perfect digital entertainment—or about an insatiable corporation’s desire to commandeer eyeballs. Instead, she said that “SKAM” had begun at the Norwegian public-television network NRK, which is essentially the BBC of Norway, where she was working in the children’s division. Before making a show, producers in the division conduct in-depth interviews of their target audience. “They try to find the need, and then they make something to meet that need,” Andem explained. The technique had been pioneered in Silicon Valley, to help techies figure out what devices were missing from our lives.

NRK had noticed that its programming wasn’t reaching older teen girls. “It had lost them to YouTube and Netflix,” Marianne Furevold-Boland, an NRK executive, told me. So the network asked Andem and Magnus to talk with Norwegian girls between the ages of sixteen and eighteen and find out what they longed to watch. Within eight months, Andem and Magnus had amassed several hundred interviews, and identified the need for a show that helped teens feel less overwhelmed and isolated. They thought that it would be fortifying for teen-agers to witness fictional young people navigating the treacherous waters of social life and social media—and surviving them.

The Internet component of “SKAM,” Andem said, had effectively been repurposed from shows for preteens that she had helped develop for NRK. But the audiences of those shows had been too young to participate fully in an online realm. (Officially, Facebook and Instagram are off limits to users younger than thirteen.) The older audience for “SKAM” could effortlessly integrate the show into the unfurling drama of their online lives.

Andem and her colleagues knew that teens spent time on the Internet, in part, because they could discover things there that their parents didn’t want them to see. So when “SKAM” débuted on NRK’s Web site, in September, 2015, it arrived without advertising or publicity. “We were terrified they would hear their mothers say that NRK had recently made an awesome show for young people,” Andem explained to Rushprint, a Norwegian film magazine. The ploy worked: teens found “SKAM” by word of mouth. By the end of Season 2, ninety-eight per cent of Norwegian teens between fifteen and nineteen knew about the show—more than knew about “Game of Thrones.”

It helped that “SKAM” was a fast-paced, sexually explicit drama about the turbulent lives of affluent sixteen-year-olds at an Oslo high school, and that it dealt with pivotal issues in teen life: coming out, sexual assault, ethnic discrimination. But the true secret of the show’s success was that it was mostly about how it feels to be in high school—when a social gaffe feels like the end of the world, and a first kiss feels like the start of a new one. As Andem puts it, “Everything’s exciting and scary.” She captures this intensity with her heightened filming style: claustrophobic closeups of teens arguing on video chat; streaky, slow-motion pans of friends dancing at a party.

In Austin, Andem had been researching teen life in Texas, trying “to understand why Americans are the way they are.” (Most Norwegian teens, Magnus noted, simply go to school and go home, whereas American teens are endlessly involved in after-school activities.) As in Norway, the Austin story lines had been shaped, to some extent, by conversations with teens. In an attempt to find nonprofessional actors, “SKAM Austin” had scouted talent at local skate parks and high schools. Fourteen hundred kids showed up for an audition at the casting agent’s office, and Andem saw half of them herself. She warmed the candidates up with improv games. “Everything they improvise yields information about who they are,” she pointed out. “That’s part of the study of who American teens are.” She favors teens who volunteer their thoughts on the script. If someone tells her, “I wouldn’t say this line,” she changes it.

Andem wants the dialogue on “SKAM” to feel raw and unscripted. She films rehearsals, because a less polished take often strikes her as the best. And she is excited by the dramatic novelties of the multi-platform format. She spoke of a moment in the middle of the Norwegian show’s second season, when an ethereal young woman named Noora was waiting for a call or a text from William, a young man with whom she was having a relationship. They had had a fight, and she hadn’t heard from him since. This was fairly conventional dramatic material, but with scenes being posted in real time, Andem said, Noora’s predicament felt agonizing. In an era of instant gratification and total information, frustration turns out to be one of the most powerful sources of drama. For about a week, William kept silent—no clip dropped. Andem recalled that other employees in the NRK offices were “just sitting there, refreshing the ‘SKAM’ page” on their computer screens. She and Magnus took further advantage of the moment after noticing, in the comments section of the show’s Web site, a young woman’s lament: “I can’t concentrate on my exam until William has answered.” Andem and Magnus transferred these words to one of Noora’s friends, who typed them during a group chat with Noora. Scripted drama had morphed into real drama, and then morphed back into fictional drama.

SKAM” ran for four seasons, and became a worldwide phenomenon. Four thousand fan fictions were written about the characters. France, Germany, and Italy produced their own versions of the show. On Weibo, the Chinese counterpart to YouTube, subtitled clips of the Norwegian “SKAM” were viewed a hundred and eighty million times.

Traditionally, the television screen has not been something that you communicate with; it’s like a professor lecturing. Your smartphone is a friend who has your ear. You gossip, plan, and hang out with it. It is axiomatic that the way we tell stories changes as new technology emerges; the rise of the novel would have been impossible without cheap paper and movable type. But it’s also true that a story is responsive to the environment in which it’s told. Ghost stories gain energy from lambent campfire; a romantic kiss becomes more intense when it is flickering on the gigantic screen of a darkened movie theatre.

Almost since the start of the smartphone era, film and TV producers have been trying to figure out how to capitalize on our new habit of jumping from one screen to the next. At first, many of these efforts felt like tricks. In 2006, a video blog called lonelygirl15 featured an ordinary-seeming teen-ager who posted regular updates about her life on YouTube and interacted with her fans on her MySpace page. The teen-ager was later revealed to be an actress; the events were fictional. In 2000, “Big Brother,” a reality show on CBS, in which roommates conspire against one another, was supplemented with streaming footage of the contestants, but it seemed to be an afterthought, like the outtakes included on the DVD of a film.

With “SKAM,” the multi-platform approach feels organic—after all, the characters themselves are constantly shuttling among YouTube and Instagram and Facebook Messenger. A teen-age “SKAM” fan named Daniel Mo was at first mystified by the show’s structural complexity, given that its story lines could have been told the old-fashioned way. Mo said, “I remember asking myself, ‘Is this really necessary?’ And the answer is yes.” One day, he realized that he was giving “likes” to posts by “SKAM” characters, just as he did to posts by close friends. Because “SKAM” flowed seamlessly into his social-media accounts, his sudden awareness of a character’s troubles often caught him off guard, and he was genuinely moved.

Mo responded to my question on a Wednesday at 10 P.M.—a time when teen-agers tend to be on their phones. “SKAM Austin” was thirteen days old, and twelve scenes had dropped, which amounted to about seventy-six minutes of footage. The audience had met the four girls who, along with Megan, formed the core of the ensemble, and had watched them flirt, quarrel, hug, and dis. (Sarah Heyward, a television writer who worked on “Girls,” had been collaborating with Andem on the scripts.) The characters spent a lot of time with their noses nearly touching portable screens, trying to make sense of their world, which is exactly what viewers were doing by following them. The Internet has a possessive imperative—you want to grab what you see before it disappears—and many “SKAM” posters had aligned themselves with particular characters, as if choosing sides in a football game. One chose Kelsey. A second wrote, “Megan totally represented me when hot guys walk in front of me.” Another declared, “Jo’s still my fave.” A fourth announced, “Grace is my current mood.”

Andem had told me that she enjoys watching soap operas, and I suspected that the ugly personal history between Megan and Abby would not be forgotten. I wasn’t disappointed. A further interrogation of Megan’s Instagram account revealed that, on New Year’s Day, 2018, she had posted an image of a sunset captioned with the words “They say time heals all wounds but how can it when you’re so hurt”; a post nine days later promised, “This is going to be my year.”

After “SKAM Austin” launched, forty-one Instagram posts by Megan became public. A selfie that she had taken in front of a mirror included, on a wall in the background, an old photograph of her in a dance leotard. I was initially confused by the post’s date—October 10, 2017—because Facebook Watch didn’t announce that it had acquired “SKAM” until about a week afterward. Looking further, I could see that the two girls’ accounts included posts that had supposedly appeared in the summer of 2016. I thought about how thrilled Magnus and Andem must have been when they realized that, because Facebook owns Instagram, “SKAM characters could now have fake Instagram histories that went back years.

In a scene that dropped on the day that Megan’s account became public, she video-chatted with a friend who had been with Marlon at another schoolmate’s house. “We left hours ago,” the friend told Megan. Later, Megan asked Marlon where he’d been, and he claimed that he’d just left the schoolmate’s house. Megan suspected that Marlon was secretly hooking up with Abby. So did viewers. “Anyone else think Marlon is cheating?” one poster asked, garnering fifty-four likes and thirty-four comments.

Fans soon noticed that, on Marlon’s Instagram account, a comment from Abby had appeared at the bottom of one of his posts: “CALL ME.” What did this mean? Screenshots of Abby’s comment spread across social media.

Minutes later, another clip appeared on Facebook Watch, which showed Megan opening her Instagram feed and clicking on the post from Marlon. She saw Abby’s comment and did a double take. She anxiously looked through Abby’s Instagram account, then tried to call Marlon, but was sent to voice mail. She returned to Marlon’s account. Abby’s comment had vanished! Who had deleted it? Abby or Marlon was the only possibility. The clip closed in on Megan’s face: you could see her drawing the same conclusion.

Viewers checked the fake Instagram account en masse, and discovered that Abby’s comment had indeed disappeared.

Ideally, the “SKAM” viewer experienced this sequence on two screens—one opened to Facebook and the other to Instagram. It was a bit of drama that seemed designed expressly for digital savants. Some viewers had clearly been left behind. “Why’d you delete @abby_taffy’s comment?” one poster asked, receiving thirty-three likes. A poster named drake.301 asked if Megan and Marlon were real. Another poster explained to him that “SKAM” was a show. Drake.301 said he knew that, but he seemed to think that he was watching reality TV. “Are they really a couple?” he asked. A user named its_ayliin set him straight: “These accounts & posts are only for the purpose of the show, they aren’t real life.”

For people who find the digital hopscotch of “SKAM” too frenetic, the clips are packaged into compilations at the end of each week. More closely resembling ordinary TV “episodes,” they include credits, theme music, and Facebook Watch’s logo. Within two and a half weeks of the launch of “SKAM Austin,” the first compilation had accumulated 7.4 million views. Individual clips were averaging around a hundred and fifty thousand views. These numbers seemed impressive—recently, the season première of “Riverdale” attracted only 2.3 million viewers—but they may be misleading, since Facebook defines a “view” as someone looking at a video for at least three seconds. Facebook can easily tabulate how many viewers are watching an entire clip and how many are quickly clicking away, but it guards such information closely. I kept asking for these numbers, but Facebook executives declined to provide them.

During the show’s second week, I met with its social-strategy manager, Michael Hoffman, who, with a razor-fade haircut and joggers, looked young enough to be Marlon’s best friend. I had the impression that an online fan community for “SKAM” had emerged spontaneously, but Hoffman told me that he had carefully guided the process, in part by creating Facebook groups and Instagram pages to encourage interactivity. Facebook Watch, I learned, had generated some of the gifs on the Instagram fan page; a young female fan on Instagram, who had posted a photograph of herself with “SKAM” scrawled across her chest in hot-pink lipstick, was a paid “influencer.”

Most fans didn’t seem to be bothered by such tactics—the influencer’s photograph received twelve hundred likes on the “SKAM” fan page. The Instagram page of Pameluft, another paid influencer, noted that her posts about “SKAM” were “sponsored,” yet commenters treated her as just another fan: “yes i love SKAM too omg,” a poster called N.UEaO wrote. Hoffman told me, “It’s about injecting our work into the right places, seamlessly.”

The show is structurally so dazzling that it’s possible to overlook the fact that it also represents an advance in invasive corporate entertainment. During the week of Kendrick Lamar’s concert, his songs accompanied one slow-motion shot after another, and Megan’s Instagram account posted a photograph of Marlon with the caption “DAMN.”—the title of Lamar’s 2017 album. Like so many Internet creations, “SKAM” seems liberatory in its cleverness, but, like the latest killer app, its ultimate purpose is to make money.

Andem acknowledged that “SKAM” was trying to manipulate viewers for maximum engagement, but she has insisted that she is not making it in order to become rich. Her aim, she said, is to help American teens feel less alone. “I think that it’s maybe more important for the teens here, because it feels like they are even more dependent than Norwegian teens,” she told me. Since moving to Texas, she said, she had been surprised to discover how much time American teens spend with their parents.

True to its roots in public television, “SKAM” attempts to educate its audience, and its primary theme is that, if you keep trying, things will come out all right in the end. In the Norwegian version, the girl who is slut-shamed for kissing someone else’s boyfriend faces down her tormentors. A young man who attempts to suppress his homosexuality winds up accepting himself. And since the U.S. version seems to echo most of the Norwegian show’s broad plot points—as did iterations in France, Germany, and Italy—something similar is likely to happen in Austin. Predicting how much “SKAM Austin” will deviate from the original is a major source of engagement on fan sites, but, whatever the variations, the show’s message will be the same: shame is transitory; growth is lasting. “Teen-agers need to build their self-esteem so that they are capable of being their own individuals, and making decisions on their own,” Andem said. “And ‘SKAM’ inspires young people to do that.” ♦

Mino Games Is Hiring Programmers in Montreal

$
0
0

Mino Games is a mobile gaming studio that strives to build the best possible games. We produced the hit mobile game Mino Monsters, with over 15 million downloads. Our development studio is in Montreal, and we are building one of the top teams in the industry.

We are funded by the leading angel, institutional investors, and gaming companies from across the world. (Andreessen Horowitz, Y Combinator, Sybo Games).

We are rapidly growing our Montreal studio, and looking for world class talent to come join us.

WeWork Is Raising Funds at $35B Valuation

$
0
0

WeWork Cos. is seeking to raise funds at a $35 billion valuation, a price tag that would place the co-working startup above companies like Airbnb and SpaceX, according to an executive at SoftBank Group Corp., which is a major WeWork investor.

Rajeev Misra, who runs SoftBank’s $100 billion Vision Fund and is the chief executive officer of SoftBank Investment Advisors, said Tuesday at the CogX conference in London that even though WeWork has been criticized as overvalued at $17 billion, it’s now raising money at $35 billion, according to a person who saw a video of the talk, which is no longer available online. Misra also said that WeWork could at some point be a $100 billion company, but didn’t comment further on whether SoftBank would participate in this funding. WeWork declined to comment.

WeWork, which runs shared office space for startups as well as large enterprises, has raised billions of dollars in equity, including $4.4 billion last summer from SoftBank, which split the money between WeWork’s central business and its three Asian subsidiaries. Misra’s comments were reported earlier by Business Insider.

For more cash, WeWork tapped the bond markets in April when it sold $702 million in seven-year unsecured bonds. The company’s bond offering documents showed fast-growing sales but even faster-increasing losses. WeWork had total revenue of $886 million but a net loss of $933 million in 2017, according to the documents. It has also committed to pay $18 billion in rent in the coming years for the buildings it currently leases, $5 billion of which will be due in the next four years.

Stick and Rudder: An Explanation of the Art of Flying (1944)

$
0
0

IdentifierStickAndRudderAnExplanationOfTheArtOfFlying

Identifier-arkark:/13960/t5s81v17m

OcrABBYY FineReader 11.0 (Extended OCR)

Ppi300

ScannerInternet Archive HTML5 Uploader 1.6.3

Elon Musk’s Boring Co. Wins Chicago Airport High-Speed Train Bid

$
0
0

Elon Musk's Boring Co. is the winner in a bid to build a multibillion-dollar high-speed express train to Chicago’s O'Hare International Airport, according to people with knowledge of the matter. The result gives the young company a big boost in legitimacy as it tries to get transportation projects underway in Los Angeles and Washington, D.C.

The company beat out a consortium that included Mott MacDonald, the civil engineering firm that designed a terminal at London's Heathrow Airport, and JLC Infrastructure, an infrastructure fund backed by former basketball star Earvin “Magic” Johnson, said the people, who asked not to be identified because they weren’t authorized to speak publicly. The city is expected to announce the news as soon as Thursday, one person said.

It’s a sizable victory for a company that was launched just 18 months ago, is working with unproven futuristic ideas, and—aside from a test tunnel it is digging in the Los Angeles suburb Hawthorne, California—lacks construction experience.

“Elon Musk is looking for a place to prove his technology works, and Chicago is rolling out the red carpet for him,” said Joe Schwieterman, director of the Chaddick Institute for Metropolitan Development at DePaul University in Chicago.

It is unclear exactly what the Boring Co. high-speed airport link would involve, but last year Musk tweeted about his ideas for Chicago. "Electric pods for sure," he wrote. "Rails maybe, maybe not."

The project is unusual in that no government funding is involved, forcing the winner to finance the entire construction cost itself. That limited interest from bidders, and has caused some to take a skeptical view of the so-called O'Hare Express project, which has been in the works for years.

"I suspect it's going to evolve a few times before anything concrete gets done," said Hani Mahmassani, a professor of engineering at Northwestern University. That said, with his other companies, Musk has "been able to pursue and fulfill visions that others say are too difficult," Mahmassani added.

Winning the nod means the city of Chicago will negotiate exclusively with Boring for one year over details of the project, which aims to connect downtown Chicago with O'Hare, about 15 miles and a $40 taxi ride away. A final go-ahead requires approval from the city council.

In its request for proposals, the city set a goal of connecting downtown with the airport in 20 minutes or less, with service every 15 minutes for the majority of the day. It also requested that fares be below the current rates for taxis and ride-share trips.

Currently, Chicagoans can ride to the airport for $5 on a Chicago Transit Authority train, taking about 40 minutes.

Viewing all 25817 articles
Browse latest View live


<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>