Start Up No.1211: Google/YouTube moderators speak out, the adaptive UI, Facebook goes DeepText, on Corbyn in 2015, and more


Guess what, Marissa Mayer’s back – and she wants to save you some time. CC-licensed photo by TechCrunch 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. Never garbage in, only out. I’m @charlesarthur on Twitter. Observations and links welcome.

Google and YouTube moderators speak out on the work that’s giving them PTSD • The Verge

Casey Newton:

»

Peter is one of hundreds of moderators at the Austin site. YouTube sorts the work for him and his colleagues into various queues, which the company says allows moderators to build expertise around its policies. There’s a copyright queue, a hate and harassment queue, and an “adult” queue for porn.

Peter works what is known internally as the “VE queue,” which stands for violent extremism. It is some of the grimmest work to be done at Alphabet. And like all content moderation jobs that involve daily exposure to violence and abuse, it has had serious and long-lasting consequences for the people doing the work.

In the past year, Peter has seen one of his co-workers collapse at work in distress, so burdened by the videos he had seen that he took two months of unpaid leave from work. Another co-worker, wracked with anxiety and depression caused by the job, neglected his diet so badly that he had to be hospitalized for an acute vitamin deficiency.

Peter, who has done this job for nearly two years, worries about the toll that the job is taking on his mental health. His family has repeatedly urged him to quit. But he worries that he will not be able to find another job that pays as well as this one does: $18.50 an hour, or about $37,000 a year.

«

People paying the price of all the other people.
unique link to this extract


Marissa Mayer is launching a new project: Lumi Labs • Fast Company

Harry McCracken:

»

Mayer remains cryptic about the specific types of apps Lumi has under development, and the time frame for their launch. But she will say that Lumi stands to benefit from the kinds of AI breakthroughs that Silicon Valley researchers are making in areas such as teaching cars to drive themselves. This kind of work, she says, is immediately useful for the tools Lumi is devising to automate activities “so mundane and so time-consuming that a lot of people [choose not to] do them.” For instance, the company is applying machine learning to certain photo-related tasks such as figuring out whether a particular image “is blurry, whether it’s well lit, whether it’s one that someone is likely to want to share based on the history of photos they shared in the past.”

If Lumi’s apps take off, it won’t be through the company’s use of AI alone. “We want our products to be thoughtful, to feel nice when they’re used,” explains Mayer, who was once famous for zealously guarding Google’s search engine against complication and clutter. She admits that she misses the days when the products she launched reached hundreds of millions of people. But with Lumi, “the hope is to be able to have that kind of impact and scale at some point,” she says. “That’s certainly what we will be building for.”

«

Whether my photo is too blurry. Really. Too blurry. From the woman who once oversaw Google Mail.

Let’s check back in two years.
unique link to this extract


How AI will eat UI

Artyom Avanesov:

»

When AR wearables hit the market, our apps will start tracking both our conscious and subconscious behavior. By measuring our heart rate, respiration, pupil size, and eye movement, our AI’s will be able to map our psychology in high resolution. And armed with this information, our interfaces will morph and adapt to our mood as we go about our day.

Future interfaces will not be curated, but tailored to fulfill our subconscious needs. Maybe the best way to navigate a digital ecosystem isn’t through buttons and sliders. Maybe the solution is something more organic and abstract.

Autodesk is developing a system that uses Generative Design to create 3D models. You enter your requirements, and the system spits out a solution. The method has already produced drones, airplane parts, and hot rods. So it’s only a matter of time before we start seeing AI-generated interfaces.

This may all sounds far out, but the future tends to arrive sooner than we expect. One day, in a brave new world, we will look at contemporary interfaces the same way we look at an old typewriter; gawking at its crudeness and appreciating how far we’ve come.

«

Now that’s something to think about. What if the UI is different for each of us because the AI picks up different things? Nobody’s phone would look the same, nobody’s phone would act the same. You wouldn’t be able to make sense of your best friend’s device. And yet it might happen.
unique link to this extract


Google Turkey suspends services for upcoming phones over fine • Daily Sabah

:

»

Tech giant Google has suspended its services for new Android smartphones in Turkey unless the country backtracks from its decision to fine the company for violating competition law, the company announced Sunday.

The decision will not affect current users or current phone models already existing on the market. The move will only suspend Google services for Android devices yet to be released.

Turkey’s Competition Authority last September announced it had fined Google some TL 93 million for violating competition laws with its mobile software sales. The watchdog said in March this year that it was launching a broader investigation into Google based on preliminary findings.

Google told Turkish business partners, phone manufacturers and telecom carriers selling smartphones that it would not grant licenses to Android phones set to be launched on the Turkish market for the use of its services, including Google Play Store, Gmail, YouTube and other Google applications. Accordingly, Google said it would also suspend operating system updates.

…The initial probe aimed to determine whether Google’s contracts with equipment producers – in addition to its mobile communications systems, applications and provision of services – found the tech giant had violated the law.

«

That’s quite the reaction, Google. Most of the 10m smartphones sold in Turkey annually run Android. Wonder how this is going to pan out if neither side backs down.
unique link to this extract


Introducing DeepText: Facebook’s text understanding engine • Facebook Engineering

Ahmad Abdulkader, Aparna Lakshmiratan, and Joy Zhang:

»

DeepText is already being tested on some Facebook experiences. In the case of Messenger, for example, DeepText is used by the AML Conversation Understanding team to get a better understanding of when someone might want to go somewhere. It’s used for intent detection, which helps realize that a person is not looking for a taxi when he or she says something like, “I just came out of the taxi,” as opposed to “I need a ride.”

We’re also beginning to use high-accuracy, multi-language DeepText models to help people find the right tools for their purpose. For example, someone could write a post that says, “I would like to sell my old bike for $200, anyone interested?” DeepText would be able to detect that the post is about selling something, extract the meaningful information such as the object being sold and its price, and prompt the seller to use existing tools that make these transactions easier through Facebook.

DeepText has the potential to further improve Facebook experiences by understanding posts better to extract intent, sentiment, and entities (e.g., people, places, events), using mixed content signals like text and images, and automating the removal of objectionable content like spam. Many celebrities and public figures use Facebook to start conversations with the public. These conversations often draw hundreds or even thousands of comments. Finding the most relevant comments in multiple languages while maintaining comment quality is currently a challenge. One additional challenge that DeepText may be able to address is surfacing the most relevant or high-quality comments.

«

But fake news? Perish the thought.
unique link to this extract


September 2015: Last house on the Left: following Jeremy Corbyn’s campaign trail • The Quietus

Taylor Parkes, back in September 2015, when it looked as though Corbyn was going to be elected leader of the Labour Party:

»

The fact is, unless a lot of things change deeply and most unexpectedly over the next four years, Jeremy Corbyn is not going to win a general election. This is not to suggest that there’s some kind of objective, immovable “centre ground”, nor that if there were, it would be occupied by the Labour Right – still less the modern Conservative Party. In truth, Corbyn’s domestic policies are not very extreme, and would in many cases prove quite popular. Yes, they’re “radical” in the sense that there’s a chasmal distance out to there from where we are today, but really, Corbynism is just about hauling Britain back towards the social-democratic Centre. There will be no pogroms, no fifteen-hour queues for stale bread. This is not the problem.

I think we all know what the problems are. For instance, I’m not what you’d call a hawk, but please: out there in grainy, hard-bollocked reality, Corbyn’s foreign policy would not just leave Britain naked in the conference chamber, but fastened into a gimp mask with a horse-tail dangling out of its arse. Whether we like it or not, there is at least one confrontation coming; you can be sure of that. There are some nasty people in the world, you know. Some of them – get this! – are even nastier than Tony Blair. And even if you leave them all alone, they will not stop. Not for all the tea in Islington North.

What’s more, there are certain… issues with Corbyn and the company he keeps. He doesn’t just have skeletons in his closet, he hangs up his shirts in an ossuary. This is not a trivial matter. Those who underestimate the problems this will cause are fooling themselves (and in some cases, losing sight of their own moral compass).

«

It’s amazing: Parkes gets every single thing correct about Corbyn, about his outriders, and his past, about how he would fare against Boris Johnson. Four. Years. Ago.

It’s a fantastic piece; I highly recommend all of it.
unique link to this extract


Post Office coughs £57.75m to settle wonky Horizon IT system case • The Register

John Oates:

»

The UK’s Post Office has finally agreed to settle a long-running case brought by postmasters the company accused of theft based on evidence from the Horizon IT system.

Claimants (and their lawyers, of course) will split £57.75m in order to settle Bates and others v the Post Office.

The biz said in a statement: “The Post Office would like to express its gratitude to claimants, and particularly those who attended the mediation in person to share their experiences with us, for holding us to account in circumstances where, in the past, we have fallen short and we apologise to those affected.”

It said the new chief executive was committed to learning lessons and that the company would be “undertaking an ambitious and sustained programme of changes to the Post Office’s relationship with postmasters”.

Freelance journalist Nick Wallis, who has been reporting on the case since 2010, pointed out that litigants would have spent about £22m, assuming their legal bills were similar to the Post Office’s. Wallis noted the case was backed by litigation funder Therium and by his rough maths on what they would expect to be paid, he estimated payments for each of the 550 litigants would be between £47,000 and £78,000.

«

Astonishing that something like this has to go on for so long, and yet the consequences for those who behaved wrongly will, one fears, be minimal.
unique link to this extract


FixMyStreet for TfL — now live • mySociety

Myfanwy Nixon:

»

Back in November, we announced our new partnership with Transport for London. We’re now pleased to say that the new Street Care service is live.

If you’re a seasoned user of FixMyStreet, there’s no learning curve required: you can proceed exactly as normal. If you prefer, you can carry on making reports through the national website at FixMyStreet.com or via the FixMyStreet app.

The only difference is that now, if the issue is the responsibility of TfL, that’s where your report will be routed, and that’s where updates will come from to let you know when the fix is in progress or completed.

The new service covers potholes, roadworks, bus shelters and traffic lights on the capital’s busiest roads — the ‘red routes’, which make up only 5% of the city’s highways, but account for a whopping 30% of traffic. Users can also report graffiti and flyposting, problems with hoardings, scaffolding and mobile cranes, street lights and damaged trees.

As ever, the underlying FixMyStreet platform means that you don’t need to think about who is responsible for your issue. If a problem is reported and it’s nothing to do with TfL, it’ll be automatically routed to the relevant borough or authority.

«

Amazingly, FixMyStreet dates back to 2007 – it’s one of the earliest web projects built for the community in the UK. Only slightly concerning that it has taken 12 years for the capital’s transport authority to integrate it.
unique link to this extract


Netflix’s Napster moment • Tech-Thoughts

Sameer Singh:

»

All of a sudden, Netflix finds itself in a world where must-watch content is fragmented across streaming services with individual subscriptions. Not that different from the pre-cord cutting world. The content economics are largely similar and all that has changed is that it is now delivered over the internet. In this world, does Netflix have the freedom to cut down on original content investments? The subscribers they have acquired have their own niche tastes. If they no longer get new content  they are interested in, they have enough options available from competitors.

Even if Netflix continues to invest in content, is that enough to keep subscribers around in this world? Users are unlikely to sign up AND stick to every streaming service that has a “must watch” show. A fan of Stranger Things, His Dark Materials and The Mandalorian is unlikely to pay for Netflix, HBO Max and Disney+ every month. A more likely outcome is that users “hop” between streaming subscriptions based on what they want to watch. There is already evidence of this, as HBO NOW subscriptions in the past few years experienced a dramatic peak during every new season of Game of Thrones. Subscriptions then dropped back down as soon as the season was complete. This pattern will dramatically increase user churn and, consequently, customer acquisition costs. Of course, the other eventuality is an increase in piracy, which will also hurt economics.

…The upside is that Netflix isn’t the only company facing these challenges. The entire video streaming industry is on an unsustainable path. High content costs, subscriber churn and piracy will affect everyone in the industry. This, in turn, is likely to create the conditions necessary for a new industry structure. Take yourself back to the music industry in the early 2000s. Revenue losses caused by piracy, and Napster in particular, forced industry players to co-operate and created the conditions necessary for Apple to unbundle music albums via iTunes. Video streaming is on a very similar path (a combination of unbundling and re-aggregation onto a single platform).

«

Singh is always worth listening to – though I think it’s no secret that this splurge on video content can’t last.
unique link to this extract


Prime leverage: how Amazon wields power in the technology world • The New York Times

Daisuke Wakabayashi:

»

The A.W.S. [Amazon Web Services] database service, an instant hit with customers, did not run software that Amazon created. Instead, the company plucked from a freely shared option known as open source.

…open source is a tried and true model nurtured by the software industry to get technology to customers quickly. A community of enthusiasts often springs up around the shareable technology, contributing improvements and spreading the word about its benefits. Traditionally, open-source companies later earn money for customer support or from paid add-ons.

Technologists initially paid little attention to what Amazon had done with database software. Then in 2015, Amazon repeated the maneuver by copying Elasticsearch and offering its competing service.

This time, heads turned.

“There was a company that built a business around an open-source product that people like using and, suddenly, they have a competitor using their own stuff against them,” said Todd Persen, who started a non-open-source software company this year so there was “zero chance” that Amazon could lift his creations. His previous start-up, InfluxDB, was open source.

Again and again, the open-source software industry became a well that Amazon turned to. When it copied and integrated that software into A.W.S., it didn’t need permission or have to pay the start-ups for their work, creating a deterrent for people to innovate.

That left little recourse for many of these companies, which could not suddenly start charging money for what was free software. Some instead changed the rules around how their wares could be used, restricting Amazon and others who want to turn what they have created into a paid service.

«

This piece begins promisingly – OMG Amazon totes ripped off Elasticsearch! – but you gradually realise that the complaints are nothingburgers. Open source companies are whining because Amazon is using the combination of its size and software that is provided as open source to produce big services they can’t compete with because they’re small.

Clue for you, people: don’t make it open source. Do it the hard way: closed source, and find customers. It worked for Microsoft and for Amazon and a gazillion companies up and down the chain. (Thanks Nic for the link.)
unique link to this extract


Errata, corrigenda and ai no corrida: none notified

6 thoughts on “Start Up No.1211: Google/YouTube moderators speak out, the adaptive UI, Facebook goes DeepText, on Corbyn in 2015, and more

  1. Like, Clippy, but for the whole UI, and replacing the basic UI instead of being superimposed on it ? Who wouldn’t love that ?

    Also, how to solve the inobvious-UI problem by making it even less obvious. I used to tell “my” devs to think of their moms trying to use the apps/sites to try and instill some awareness and empathy. Now it’ll be “tell that to the UIAI team !”

    And finally, tech training/support is social. IT is a language; differentiating by user is Babel all over again.

    They should probably start by trying to design an independent smart UI help system, so that users get confident asking the app for help on itself, instead of me ^^

  2. I just discovered Wikipedia has a list of the best-selling smartphones (not by OEM, by individual model): https://en.wikipedia.org/wiki/List_of_best-selling_mobile_phones#2019

    Kind of interesting/surprising. Of course this varies wildly by country/region because especially the new Chinese OEMs have very patchy/uneven coverage, but still:
    – I’m surprised Huawei sells more flagship P/Mate than midrangers/Honor handsets
    – It’s encouraging Redmi sold more Notes than 6A. 6A is a very cut-down device (no ARM A7x, no touchID, usually 2GB/16GB), a clear case of paying a little but less to get a whole lot less.
    – Samsung’s midrange is strong.

    It’s a pity media coverage focuses on flagships. There are precious few resources to evaluate low/mid-range handsets, especially in the Western world. I go with Techtablets.com and gsmarena.com.

    • Oh, and I just realized Android Go, the lighter version of Android, has vanished from the radar. It’s still around, but no announcement and the “lite” version of apps that appeared after it was launched have trickled to nothing, so slow death.
      It never made a whole lot of sense, the basic SW was the same, only optimized for slightly less CPU/storage/data, so a whole lot of ongoing effort for $20 of hardware, and for low-value customers. Google got its tentacles into KaiOS anyway so they’ve got that low-end monetized w/o the headache of being in charge of a discount ecosystem.

  3. Re Google and Turkey: the reporting is bad on that one. Engadget says “change its software distribution agreements in order to allow consumers to choose a different search engine for their Android devices.”. But that’s always been possible: by changing the browser default search engine, by using a different browser, and/or by using a different search app+widget, assistant…

    Is this about defaults and a getting a setup-time voting screen like the EU got in Windows for Browsers, or am I missing something ?

  4. This one is interesting: large scale blind test of smartphone cameras over social media: https://www.youtube.com/watch?time_continue=353&v=KxsFat1ImiY&feature=emb_logo

    Surprisingly, Samsung gets the two top slots, with a rather under-hyped and staid camera subsystem.

    Mostly, it’s a valid test because a) it’s blind which is ground zero for a valid test and b) it’s over social media, where I’d guess 90% of real-world pictures end up.

    If one wants to nitpick, a couple of issues:
    – it’s by direct elimination, which makes results for all but the top 2 phones unfathomable.For example, we’ll never know how the Mi Note 10 compares to a Pixel 4 because they didn’t get matched. And the Mi Note 10 lost in the first round, but to a finalist, so there’s no knowing whether it’s better than any of the other phones that made it past 1/2/3 rounds.
    – there was only one pic, and it changed for each round. So there was an element of luck about a phone getting a pic that is its opponent strong or weak point. Apparently iPhones have an issue with white balance while OnePlus doesn’t, and the first pic put that into play, so the iPhone got dumped on the first pic.
    – it was a choice of likeable picture by regular Joes, not an evaluation of best technical pic by experts. What looks pleasing vs what iss accurate, rich, a good basis for some post-prod work. That’s fair though, most people want a cute pic, not a post-prod-ready one.

Leave a comment

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