Start Up No.1,172: facial recognition v your job, background monk noise, how pollen lowers crime, Gmail’s costly fill, is AR gaming the answer?, and more

Here’s an odd thing: certain US political TV ads aren’t allowed to lie. Et tu, Facebook? CC-licensed photo by Nicole Lee on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

A selection of 10 links for you. Use them wisely. I’m @charlesarthur on Twitter. Observations and links welcome.

HireVue’s AI face-scanning algorithm increasingly decides whether you deserve the job • The Washington Post

Drew Harwell:


An artificial intelligence hiring system has become a powerful gatekeeper for some of America’s most prominent employers, reshaping how companies assess their workforce — and how prospective employees prove their worth.

Designed by the recruiting-technology firm HireVue, the system uses candidates’ computer or cellphone cameras to analyze their facial movements, word choice and speaking voice before ranking them against other applicants based on an automatically generated “employability” score.
HireVue’s “AI-driven assessments” have become so pervasive in some industries, including hospitality and finance, that universities make special efforts to train students on how to look and speak for best results. More than 100 employers now use the system, including Hilton, Unilever and Goldman Sachs, and more than a million job seekers have been analyzed.

But some AI researchers argue the system is digital snake oil — an unfounded blend of superficial measurements and arbitrary number-crunching that is not rooted in scientific fact. Analyzing a human being like this, they argue, could end up penalizing nonnative speakers, visibly nervous interviewees or anyone else who doesn’t fit the model for look and speech.

The system, they argue, will assume a critical role in helping decide a person’s career. But they doubt it even knows what it’s looking for: Just what does the perfect employee look and sound like, anyway?

“It’s a profoundly disturbing development that we have proprietary technology that claims to differentiate between a productive worker and a worker who isn’t fit, based on their facial movements, their tone of voice, their mannerisms,” said Meredith Whittaker, a co-founder of the AI Now Institute, a research center in New York.


I’ve heard a couple of radio programs about this system, and it’s profoundly depressing. You have to act for the machine; it then makes the judgement about how well suited you are – the videos aren’t even reviewed by a human. It feels terribly wrong.
unique link to this extract


The Name Of The Rose • Background Ambience Generator

Dr. Ir. Stéphane Pigeon:


The Player consists of 10 colorized sliders, followed by a row of buttons. Sliders mimic a 10-band equalizer and are associated with strict frequencies when the generator is calibrated. Each slider then represents one octave. When the generator is not calibrated, the sliders are still ordered by increasing frequencies… when it is possible. For example, the rumble of a distant thunder will likely be associated with the first slider, and chirping insects with the last one.


Monastery bells, Gregorian voices, chapel voices, and more. Your background sound for today.
unique link to this extract


Revealing the link between peak season for allergies and crime • Journalists’ Resource

Clark Merrefield:


New research in the Journal of Health Economics reveals something unexpected about allergies: US cities experiencing unusually high pollen counts also experience lower rates of reported violent crime.

Looking at daily pollen counts across 16 US cities from 2007 to 2016 and crimes reported to local law enforcement that the FBI collects, the authors find that reported violent crime falls 4% on high-pollen days. That drop is about the level of crime reduction that would come with a 10% increase in the size of a city’s police force, the authors write. The paper adds to past research investigating how health and other shocks — like football upsets — affect crime rates…

…The authors focus on New York City to explore why more pollen might lead to less violent crime. Using detailed microdata they find the effect is largely due to reduced residential violent crime. Residential violent crime refers to violence that happens in the home — typically intimate partner and other domestic violence, according to the authors.

They turn to data from the bike-rental program Citibike as a proxy for outdoor activity. On high-pollen days — when pollen readings exceed 1,500 parts per million — Citibike activity declines 8%. That 1,500 ppm reading is the threshold for “high-pollen days” used throughout the paper. The average pollen count across cities during the period studied is about 172 ppm.

The authors write, “these results are consistent with the idea that individuals are more likely to remain in their homes on days in which pollen counts are high.” The behavioral change may be associated with run-of-the-mill allergy symptoms.

“Pollen makes people more fatigued, more tired,” Deza says. “The mechanism we are discussing in this paper is that if people are more fatigued and more lethargic, they may be less likely to be angry.”


unique link to this extract


Gmail hooked us on free storage. Now Google is making us pay • Bloomberg Quint

Gerrit De Vynck:


Google lured billions of consumers to its digital services by offering copious free cloud storage. That’s beginning to change.

The Alphabet Inc. unit has whittled down some free storage offers in recent months, while prodding more users toward a new paid cloud subscription called Google One. That’s happening as the amount of data people stash online continues to soar.

When people hit those caps, they realize they have little choice but to start paying, or risk losing access to emails, photos and personal documents. The cost isn’t excessive for most consumers, but at the scale Google operates, this could generate billions of dollars in extra revenue each year for the company. Google didn’t respond to an email seeking comment.

A big driver of the shift is Gmail. Google shook up the email business when Gmail launched in 2004 with much more free storage than rivals were providing at the time. It boosted the storage cap every couple of years, but in 2013 it stopped. People’s in-boxes kept filling up. And now that some of Google’s other free storage offers are shrinking, consumers are beginning to get nasty surprises.

“I was merrily using the account and one day I noticed I hadn’t received any email since the day before,” said Rod Adams, a nuclear energy analyst and retired naval officer. After using Gmail since 2006, he’d finally hit his 15 GB cap and Google had cut him off. Switching away from Gmail wasn’t an easy option because many of his social and business contacts reach him that way.

“I just said ‘OK, been free for a long time, now I’m paying,’” Adams said.

Other Gmail users aren’t so happy about the changes. “I am unreasonably sad about using almost all of my free google storage. Felt infinite. Please don’t make me pay! I need U gmail googledocs!,” one person tweeted in September.

Some people have tweeted panicked messages to Google in recent months as warnings about their storage limits hit.


2004 to 2019: that’s playing a hell of a long game.
unique link to this extract


WeWork’s failure is SoftBank’s day of reckoning • WIRED

Leonard Sherman:


Masayoshi Son seems to believe that the Vision Fund’s massive capital investments can be used as a weapon to convey sustainable competitive advantage, global domination, and superior returns for his chosen winners. But this thinking is profoundly flawed for three reasons.

1. The notion that one VC can exploit money to achieve sustainable competitive advantage is ludicrous on its face. In virtually every category in which SoftBank is heavily invested—real estate, ridesharing, meal delivery, freight brokerage, hotels, construction—SoftBank is facing well-capitalized and resilient competition. In a world awash in capital, none of SoftBank’s funded ventures has achieved anything close to monopoly pricing power. This marketplace reality has contributed to chronic and escalating losses across Son’s portfolio.

Bewilderingly, SoftBank itself occasionally backs direct competitors within the same business category such as Doordash and Uber Eats in the US and Didi Chuxing and Uber in Latin America. Not surprisingly, in these cases, SoftBank’s competing ventures have suffered deep losses.

2. SoftBank’s philosophy ignores the value of low-cost learning from stage-gated investing, and instead exposed blitzscaled ventures to massive risk and wasted resources. Capital constraints aren’t an inconvenient nuisance for early stage ventures. Rather, fiscal discipline encourages experimentation to optimize business performance in terms of product/market fit, technology reliability, supply chain efficiency, business process stability, and business model viability.

By often investing too much, too soon in unproven ventures, sometimes with minimal due diligence, SoftBank compels its portfolio companies to rapidly scale businesses that still have unproven or deeply flawed business models (e.g. WeWork and Uber), inadequate core business processes (e.g. Brandless, Wag) or weak defenses against competitive threats (Slack). Prematurely picking winners with massive bets heightens the risk that a company’s race for global domination winds up becoming a race to oblivion.

3. Even if weaponizing capital could promote winner-take-all outcomes, SoftBank has been investing in the wrong types of businesses to achieve its goal of profitable market dominance.


unique link to this extract


2014: On censoring political ads • CommLawBlog

Frank Montero, with a post from five years ago that seems somehow relevant given Facebook’s reluctance to touch anything political advert-y:


The Communications Act and the FCC’s rules prohibit broadcasters from censoring political candidates’ ads in any way if those ads are “uses”. In this context, a “use” is an ad, sponsored by a legally qualified candidate or the candidate’s campaign committee, that includes a recognizable likeness or image of the candidate. The candidate may be seeking a federal office or a state or local office. The ad buy may be the first one run by a candidate for that particular office, or it may be bought by a candidate taking advantage of the “equal opportunity” requirement created by the fact that the candidate’s opponent aired a “use” already.

If it’s a political “use”, broadcasters can’t touch the content…

…Because Congress prohibits broadcasters from censoring such ads, broadcasters enjoy immunity from liability arising from the content of such ads. So even if a political “use” contains, say, blatantly defamatory statements, the broadcaster cannot be held liable for any harm to the defamed individual. The recourse for a party claiming to be injured by the contents of such a political ad is to sue the candidate who produced the ad.

But there are a couple of very important caveats.

First, the “no censorship” prohibition applies only to “uses”. That still leaves a wide array of political ads which can legally be censored – or even rejected – because of their content. Ads by non-candidate third parties like PACs, labor unions, and other advocacy groups are not “uses”, even though they may resemble them in look and content.


So TV can refuse some ads. But Facebook won’t, at all, and isn’t liable in the way that TV is.
unique link to this extract


With $15m round and 100k tablets sold, reMarkable CEO wants to make tech ‘more human’ • TechCrunch

Devin Coldewey:


The reMarkable tablet is a strange device in this era of ultra-smart gadgets: A black and white screen meant for reading, writing, and sketching — and nothing more. Yet the company has sold 100,000 of the devices and now has attracted $15m in series A funding from Spark Capital.

It’s an unusual trajectory for a hardware startup exploring a nearly unoccupied market, but CEO Magnus Wanberg is confident that’s because this category of device is destined to grow in response to increasingly invasive tech. Sometimes an anti-technology trend is the tech opportunity of a lifetime.

I reviewed the reMarkable last year and compared it with its only real competition, the Sony Digital Paper Tablet. It was launched not on Kickstarter or Indiegogo but with its own independent crowdfunding campaign — and considering we’ve seen devices like this attempt such a thing and either let down or rip off their backers, that alone was a significant risk.


My feeling is that reMarkable has already hit its total addressable market. I’d also like to know how many who bought it are still using it after, say, 12 months.
unique link to this extract


Tilt Five: holographic tabletop gaming • Kickstarter



Tilt Five is a whole new way to play games, in Augmented Reality (AR), with freaking holograms!

When you slip on the Tilt Five glasses and look at the game board, a vibrant 3D world opens up to you.  Suddenly chasms seem to drop infinitely into your table, and game characters and monsters spring up from the game board.  This is gaming unlike anything you’ve experienced before.

Out of the box you’ll have access to thousands of officially licensed RPG adventures, as well as a multitude of classic tabletop, action video, puzzle, and party games.

“When we started designing the Tilt Five system our focus wasn’t on the technology for technology’s sake.  We wanted to provide an amazing gaming experience that blends the things you love about video games and board games. And we wanted it to be just as fun when playing solo, together with your friends, or even when you’re apart.” – Jeri Ellsworth, Co-Founder & CEO


I linked to this because it feels to me as though consumer “scenic” AR might be successful in this way – through glasses (yes) which you only wear for limited periods, for a specific task. It’s communal, you accept the limitations (it doesn’t have to fill the room) – it’s quite a clever way of limiting the required space.
unique link to this extract


Anonymous author of Trump ‘resistance’ op-ed to publish a tell-all book • The Washington Post

Philip Rucker:


The author of an anonymous column in the New York Times in 2018, who was identified as a senior Trump administration official acting as part of the “resistance” inside the government, has written a tell-all book to be published next month.

The book, titled, “A WARNING,” is being promoted as “an unprecedented behind-the-scenes portrait of the Trump presidency” that expands upon the Times column, which ricocheted around the world and stoked the president’s rage because of its devastating portrayal of Trump in office.

The column described Trump’s leadership style as “impetuous, adversarial, petty and ineffective,” and noted that “his impulsiveness results in half-baked, ill-informed and occasionally reckless decisions that have to be walked back.”

…“Picking up from where those first words of warning left off, this explosive book offers a shocking, firsthand account of President Trump and his record,” reads a statement about the book’s release.


Sorry, but it’s not a surprise any more. Quite a lot of us could tell long before he was foolishly elected that Trump was incompetent, bad at decision-making, and would favour nepotism over competence. All we’re interested in now is the downfall.
unique link to this extract


Improving how we calculate writer earnings (3 min read) • Medium

Emma Smith:


Instead of paying based on claps as the main signal, we will now reward stories primarily based on reading time, which we’ve seen to be a closer measure of quality and resonance with readers. To increase transparency and provide richer insight to our writers, we will also introduce new stats so it’s more clear how a story’s earnings were calculated…

…By calculating a share of member reading time, we support authors who write about unique topics and connect with loyal readers. For example, if last month a member spent 10% of their monthly reading time on your story, you will receive 10% of their share (a portion of their subscription fee).

Imagine an author writes about fly fishing. She finds an audience of fly fishing enthusiasts who subscribe to Medium primarily to read her stories, meaning she receives a strong share of reading time from each of her readers. In contrast, an author who writes about a wide variety of topics might receive smaller shares from a broader audience of readers, who also read a variety of other authors. While the generalist will often earn a lot through the first total reading time part, the fly fisher is well equipped to earn through this share part — even with a smaller audience.


The question then is, will Medium be a better way for that fly fishing writer to find people who’d pay money than a newsletter leveraging some other outlet? (Will writers be allowed to advertise their paid-for newsletter which might monetise better than Medium in their posts?)

Medium’s pivots around quite how it should monetise its readers must be exhausting for those trying to edit magazines on it. (I’ve earned from a number of articles I’ve written on Medium, but I don’t know how these changes would affect them.)

The question now will be what the technique is one employs to keep people engaged with an article. Write a book in chapter-length instalments, a la Dickens? Still, at least it’s not a pivot to video, just a pivot to.. legio?
unique link to this extract


Errata, corrigenda and ai no corrida: none notified

3 thoughts on “Start Up No.1,172: facial recognition v your job, background monk noise, how pollen lowers crime, Gmail’s costly fill, is AR gaming the answer?, and more

  1. I know it’s not type of info you’re interested in reporting, but the update situation on Android keeps steadily improving:

    1- the way they’re presenting the thing is a bit disingenuous: the userbase was still growing over this period. Percentages not just raw totals, would complete the picture
    2- Since updates are mostly a reactionary feature on Android (they’re mostly about security, it’s not so much good to have them as bad not to), the total of not-updated device is relevant too.
    3- Security updates are decoupled from OS updates. It’d be useful to have the figures for those too, not just OS updates.
    4- Amusingly, with Project Mainstream there’s now 3 things to track: OS version, security patch version, and ProjMain version. That’s work, I’m sure the press will mostly be doing scare pieces on random malware that needs root and isn’t on the PlayStore ^^

    Anyhoooo…. Things. Getting better. Slowly.

  2. Plausibly neutral Android vs iOS head-to-heads are rare. This is one:
    The guy has been doing reviews forever and is refreshingly blunt.

    Basically, the iPhone blows the Pixel out of the water in video (4k@60fps) and edges it out in general pictures; while the Pixel blows the Iphone out w/ Google Assistant vs Siri, and offer generally better software starting w/ the home page (free icons and widgets placement, better gestures).

    I’d really love to see task- or user-centered reviews. To my surprise, the TechTablet guy is doing his video editing on a Galaxy S10 rather than an iPad Pro because it’s … faster (multi-threading FTW !)

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.