Start Up No.1402: Facebook’s populist ‘advantage’, deepfake blood detection, do smartphones think?, moving buildings, and more


A tokomak: a more advanced version of this could become the world’s first self-sustaining nuclear fusion reactor. CC-licensed photo by David Chase 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 9 links for you. Not up for debate. I’m @charlesarthur on Twitter. Observations and links welcome.

Compact nuclear fusion reactor is ‘very likely to work,’ studies suggest • The New York Times

Henry Fountain:

»

Scientists developing a compact version of a nuclear fusion reactor have shown in a series of research papers that it should work, renewing hopes that the long-elusive goal of mimicking the way the sun produces energy might be achieved and eventually contribute to the fight against climate change.

Construction of a reactor, called Sparc, which is being developed by researchers at the Massachusetts Institute of Technology and a spinoff company, Commonwealth Fusion Systems, is expected to begin next spring and take three or four years, the researchers and company officials said.

Although many significant challenges remain, the company said construction would be followed by testing and, if successful, building of a power plant that could use fusion energy to generate electricity, beginning in the next decade.

This ambitious timetable is far faster than that of the world’s largest fusion-power project, a multinational effort in Southern France called ITER, for International Thermonuclear Experimental Reactor. That reactor has been under construction since 2013 and, although it is not designed to generate electricity, is expected to produce a fusion reaction by 2035.
Bob Mumgaard, Commonwealth Fusion’s chief executive and one of the company’s founders, said a goal of the Sparc project was to develop fusion in time for it to play a role in mitigating global warming. “We’re really focused on how you can get to fusion power as quickly as possible,” he said.

…Sparc would be far smaller than ITER — about the size of a tennis court, compared with a soccer field, Dr. Mumgaard said — and far less expensive than the international effort, which is officially estimated to cost about $22bn but may end up being far costlier. Commonwealth Fusion, which was founded in 2018 and has about 100 employees, has raised $200m so far, the company said.

«

Come on, come on… Fusion is always just out of reach. But maybe, maybe this time, it will change? If the papers are published, could anyone try?
unique link to this extract


The subtle effects of blood circulation can be used to detect deep fakes • IEEE Spectrum

David Schneider:

»

This work, done by two researchers at Binghamton University (Umur Aybars Ciftci and Lijun Yin) and one at Intel (Ilke Demir), was published in IEEE Transactions on Pattern Analysis and Machine Learning this past July. In an article titled, “FakeCatcher: Detection of Synthetic Portrait Videos using Biological Signals”, the authors describe software they created that takes advantage of the fact that real videos of people contain physiological signals that are not visible to the eye.

In particular, video of a person’s face contains subtle shifts in color that result from pulses in blood circulation. You might imagine that these changes would be too minute to detect merely from a video, but viewing videos that have been enhanced to exaggerate these color shifts will quickly disabuse you of that notion. This phenomenon forms the basis of a technique called photoplethysmography, or PPG for short, which can be used, for example, to monitor newborns without having to attach anything to a their very sensitive skin.

Deep fakes don’t lack such circulation-induced shifts in color, but they don’t recreate them with high fidelity.

«

So we win the war! Until the next update.
unique link to this extract


Why the right wing has a massive advantage on Facebook • POLITICO

Alex Thompson:

»

Throughout 2020, Democrats have denounced Facebook with growing ferocity as a “right wing echo chamber” with a “conservative bias” that’s giving an edge to Donald Trump in November.

But Facebook says there’s a reason why right-wing figures are driving more engagement. It’s not that its algorithm favors conservatives — the company has long maintained that its platform is neutral. Instead, the right is better at connecting with people on a visceral level, the company says.

“Right-wing populism is always more engaging,” a Facebook executive said in a recent interview with POLITICO reporters, when pressed why the pages of conservatives drive such high interactions. The person said the content speaks to “an incredibly strong, primitive emotion” by touching on such topics as “nation, protection, the other, anger, fear.”

“That was there in the [19]30’s. That’s not invented by social media — you just see those reflexes mirrored in social media, they’re not created by social media,” the executive added. “It’s why tabloids do better than the [Financial Times], and it’s also a human thing. People respond to engaging emotion much more than they do to, you know, dry coverage. …This wasn’t invented 15 years ago when Mark Zuckerberg started Facebook.”

«

“Hitler would have done great on Facebook” isn’t quite the ringing endorsement that the executive seems to think it is. Populism might be “engaging” but it’s not generally how the human race has thrived. Quite the opposite.
unique link to this extract


What is it like to be a smartphone? • ROUGH TYPE

Nick Carr:

»

You would not be able to know what it’s like to be an AI by examining the 1s and 0s of its machine code any more than you’d be able to understand your own being by examining the As, Cs, Gs, and Ts of your genetic code. A conscious computer would likely be unaware of the routines of its software — just as we’re unaware of how our DNA shapes our body and being or even of the myriad signals that zip through our nervous system every moment. An intelligent computer may perform all sorts of practical functions, including taking our inputs and supplying us with outputs, without having any awareness that it is performing those functions. Its being may lie entirely elsewhere.

The Turing test, in all its variations, would also be useless in identifying an AI. It merely tests for a machine’s ability to feign likeness with ourselves. It provides no insight into the AI’s being, which, again, could be entirely separate from its ability to trick us into sensing it is like us. The Turing test tells us about our own skills; it says nothing about the character of the artificial being.

All of this raises another possibility. It may be that we are already surrounded by AIs but have no idea that they exist. Their beingness is invisible to us, just us ours is to them. We are both objects in the same place, but as beings we inhabit different universes. Our smartphones may right now be having, to borrow Nagel’s words, “experiences fully comparable in richness of detail to our own.”

Look at your phone. You see a mere tool, there to do your bidding, and perhaps that’s the way your phone sees you, the dutiful but otherwise unremarkable robot that from time to time plugs it into an electrical socket.

«

Always a treat when Nick puts finger to keyboard. (“Is your smartphone conscious?” should be in an Oxbridge philosophy exam at some point in the near future.)
unique link to this extract


Financial Conduct Authority Perimeter Report 2019/20

Under section 3.20, “Mass marketing of high risk investments to retail consumers”:

»

Online platforms, such as search engines and social media platforms, play an increasingly significant role in communicating financial promotions to consumers. As a result, consumers are being more readily exposed to adverts, ranging from scams and promotions of high-risk investments to false or misleading adverts (falling either side of the regulatory perimeter) which, directly or indirectly, lead consumers onto paths resulting in harm. As the digital world continues to develop, the potential harms to consumers change in both nature and severity.

We think that it is important that online platform operators, like Google, bear clear legal liability for the financial promotions they pass on – at least to the same extent as traditional publishers of financial promotions; that would mean that an online publisher would have to ensure that any financial promotion which they communicate has first been approved by an authorised person or otherwise falls within the scope of an exemption in the Financial Promotions Order. We are currently considering with the Treasury the application of the financial promotions regime to these platform operators and whether we need any new powers over them. This work is relevant not just to the promotion of high risk investments but to our work to address online harms – including scams – more generally.

«

Emphasis added. This is the subject of a campaign offline, and the challenge is to get the Treasury to listen. Given that it will “just” mean that Google has to take a bit more time about checking adverts, you’d think the Treasury would be happy to listen to the FCA.
unique link to this extract


Introducing Amazon One—a new innovation to make everyday activities effortless • Amazon

»

Why did you create Amazon One?
As with everything Amazon does, we started with the customer experience and worked backwards. We solved for things that are durable and have stood the test of time but often cause friction or wasted time for customers. We wondered whether we could help improve experiences like paying at checkout, presenting a loyalty card, entering a location like a stadium, or even badging into work. So, we built Amazon One to offer just that—a quick, reliable, and secure way for people to identify themselves or authorize a transaction while moving seamlessly through their day.

Why did you pick palm recognition?
We selected palm recognition for a few important reasons. One reason was that palm recognition is considered more private than some biometric alternatives because you can’t determine a person’s identity by looking at an image of their palm. It also requires someone to make an intentional gesture by holding their palm over the device to use. And it’s contactless, which we think customers will appreciate, especially in current times. Ultimately, using a palm as a biometric identifier puts customers in control of when and where they use the service.

«

You can’t determine a person’s identity by looking at an image of their iris, either. But that doesn’t mean you’d want even a hashed version of your iris sitting on Amazon’s EC2. I don’t see the advantage over using a normal contactless payment system, except that with this one Amazon gets to tie your transactions to you, perfect you (and your credit card and mobile phone, required for signup).
unique link to this extract


Media’s failed attempt to take on the Facebook-Google “duopoly” • Axios

Sara Fischer:

»

Disney on Monday sold ad tech provider TrueX, per the Wall Street Journal, an asset it’s been looking to divest since it acquired the property through its acquisition of most of 21st Century Fox. Disney was never trying to develop a major ad tech business the same way some of its streaming rivals once were, although it does have ad tech businesses that help power its ESPN and Hulu streaming platforms, as well as its digital assets.

AT&T is exploring the potential sale of its ad-buying unit Xandr, per the Journal. Xandr was created through the acquisitions of the ad tech firm AppNexus and the merger with Time Warner. During the trial to buy Time Warner, executives argued that the deal made sense because it would help AT&T compete with Google and Facebook for ad dollars.

Verizon has written down half of its investment in its mostly ad-supported media arm, and reports suggest it is looking to offload HuffPost, which was once considered a traffic goldmine for an ad-supported business. Verizon is still investing in its advertising technology, but its business has taken a hit due to the coronavirus.

Google and Facebook still control an overwhelming percentage of the U.S. digital ad market, even though they are losing some ground to Amazon. 

Jason Kint, CEO of Digital Content Next, the trade group representing premium publishers, tweeted in July that he still expects Facebook and Google to bring in 88% of all new digital ad dollar growth this year.

«

Not far to go before it’s past 90%, then 95%.
unique link to this extract


Pandemic is far from over, experts say, despite Trump allies’ claims • The New York Times

Donald McNeil Jr:

»

In the last week, leading epidemiologists from respected institutions have, through different methods, reached the same conclusion: About 85% to 90% of the American population is still susceptible to SARS-CoV-2, the virus causing the current pandemic.

The number is important because it means that “herd immunity” — the point at which a disease stops spreading because nearly everyone in a population has contracted it — is still very far off.

The evidence came from antibody testing and from epidemiological modeling. At the request of The New York Times, three epidemiological teams last week calculated the percentage of the country that is infected. What they found runs strongly counter to a theory being promoted in influential circles that the United States has either already achieved herd immunity or is close to doing so, and that the pandemic is all but over. That conclusion would imply that businesses, schools and restaurants could safely reopen, and that masks and other distancing measures could be abandoned.

“The idea that herd immunity will happen at 10 or 20% is just nonsense,” said Dr. Christopher J.L. Murray, director of the University of Washington’s Institute for Health Metrics and Evaluation, which produced the epidemic model frequently cited during White House news briefings as the epidemic hit hard in the spring.

…More than 200,000 Americans have already died, and models estimate that if people return to old habits, such as gathering indoors without masks, more than 300,000 and possibly 400,000 could die before a vaccine is widely available.

…The immunity conferred by a common cold coronavirus appears to last a year or two, immunologists say, and then a person can catch the same cold again. Antibodies against it fade away; primed T-cells remain.

Primed T-cells may lower the odds of dying from the new, dangerous coronavirus, Dr. Crotty said, but that has not been proven. There is no evidence that they protect against becoming infected with it.
The experts who promote the theory that primed T-cells even stop infections typically are not immunologists.

«

“Typically not immunologists” is a neat way of saying “people who score highly on Dunning-Kruger”.
unique link to this extract


Swiss company moves 6,200 tonne building 60 metres • BBC News

»

A 122-year-old, 6,200 tonne building in Zurich is being moved 60 metres westward to make way for the expansion of a nearby railway.

The ambitious project, two years in the making, has involved freeing the foundations of the Machine Factory Oerlikon building before loading it onto a dedicated rail track.

«

There’s an accompanying video, which shows how remarkable this is. I’m constantly astonished at the process of moving a building because I keep wondering: how do you separate the walls from the foundations? (Cut through them, I guess.) Having done that, how do you lift all of the building at the same time so that you don’t destroy it through uneven stresses – which, given the size of this one, would quickly be catastrophic? Once that’s done, how do you unite it to the new foundations?

And in case you hadn’t noticed, it is somehow Infrastructure Week. Suggestions and links of absurd building deployment welcome. (Thanks Giuseppe for today’s.)
unique link to this extract


Errata, corrigenda and ai no corrida: none notified

1 thought on “Start Up No.1402: Facebook’s populist ‘advantage’, deepfake blood detection, do smartphones think?, moving buildings, and more

  1. Asking “Is your smartphone conscious?” strikes me as akin to asking “Is a fruit fly conscious?”. The problem is taking something we intuitively think of as a binary, a yes/no, and putting it into a context which virtually forces a gradient meaning, which is very prone to then confusing fogginess with philosophy.

    I was reminded of an old joke about the difference between cats and dogs. Here’s a version I found:

    Dogs think: “These people feed me, provide for me, give me shelter, and love me. They must be God!”

    Cats think: “These people feed me, provide for me, give me shelter and love me. I must be God!”

    Here the joke would be something like: “You think your smartphone is tool for you. But – what if your smartphone thinks you are a tool for it!”

    Is that just amusing, or profound?

Leave a Reply to Seth Finkelstein Cancel reply

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

WordPress.com Logo

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

Google photo

You are commenting using your Google 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.