DeepMind has made a significant breakthrough in forecasting protein folding – and it matters. CC-licensed photo by Enzymlogic 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 8 links for you. Neatly folded. I’m @charlesarthur on Twitter. Observations and links welcome.
‘It will change everything’: DeepMind’s AI makes gigantic leap in solving protein structures • Nature
Ewen Callaway:
»
The ability to accurately predict protein structures from their amino-acid sequence would be a huge boon to life sciences and medicine. It would vastly accelerate efforts to understand the building blocks of cells and enable quicker and more advanced drug discovery.
AlphaFold came top of the table at the last CASP — in 2018, the first year that London-based DeepMind participated. But, this year, the outfit’s deep-learning network was head-and-shoulders above other teams and, say scientists, performed so mind-bogglingly well that it could herald a revolution in biology.
“It’s a game changer,” says Andrei Lupas, an evolutionary biologist at the Max Planck Institute for Developmental Biology in Tübingen, Germany, who assessed the performance of different teams in CASP. AlphaFold has already helped him find the structure of a protein that has vexed his lab for a decade, and he expects it will alter how he works and the questions he tackles. “This will change medicine. It will change research. It will change bioengineering. It will change everything,” Lupas adds.
In some cases, AlphaFold’s structure predictions were indistinguishable from those determined using ‘gold standard’ experimental methods such as X-ray crystallography and, in recent years, cryo-electron microscopy (cryo-EM). AlphaFold might not obviate the need for these laborious and expensive methods — yet — say scientists, but the AI will make it possible to study living things in new ways.
«
It’s really hard to overstate the importance of this, though it will take years to become apparent; rather as sequencing the human genome in the late 1990s led to us being able to produce an mRNA vaccine in just two days, given its sequence (the rest of the time has been spent on production and trials). The complexity of protein folding is mindblowing: it depends on interactions between existing parts of the protein as it’s produced, which then are influenced by subsequent parts. And it all works.
unique link to this extract
‘Christmas slots went in five hours’: how online supermarket Ocado became a lockdown winner • The Guardian
Harry Wallop:
»
From the viewing platform you can watch these metal cubes endlessly whiz around, moving thousands of plastic crates as if they were playing an enormous game of chess. You occasionally sight bottles of bleach or rosé, packets of noodles and dog biscuits, before they are sent down to a lower level.
“I find it quite mesmerising, like robotic ballet,” says Mel Smith, CEO of Ocado Retail, the UK arm of the business. “The day I decided I wanted this job was when I went to [the warehouse] and thought, this is absolutely the future.”
Smith is a plain-speaking, cheery New Zealander who joined just over a year ago from Marks & Spencer. She was hired in part to oversee the potentially tricky move of Ocado taking on M&S as its main supplier, after its longstanding relationship with Waitrose turned sour. The day I visit, it is just 48 hours after Ocado switched to M&S.
A floor below the robots, Elizabeth, a personal shopper, is standing at pick station number 29. Her average “each” time – Ocado jargon for individual products – is flashed up on a screen in front of her: 6.7 seconds, beating the target of seven seconds. This is how long it takes her to reach into the crate that has flown down a shaft from the floors above, take a product – in this case a bag of Cafédirect Machu Picchu ground coffee – scan its barcode and place it in one of Ocado’s distinctive grey plastic bags.
This is what makes the online retailer so different. If you place an order for a large shop from Tesco, Sainsbury’s or any of the other supermarkets, they will usually send a worker to walk along the aisles and pick your shopping in either an actual store or a so-called “dark store”, which caters exclusively to online shopping. Even if they are very quick, this is likely to take 20 minutes to half an hour. “There’s no way you can pick as quickly in store as here. Nothing like it,” says Simon Nottage, who runs the warehouse and is showing me around. He points to Elizabeth, who is now placing some cat litter into a bag: “This one station will do up to 400 ‘eaches’ an hour.”
«
The start of lockdown in March was like a DDOS attack. But in a good way.
unique link to this extract
Economists urge BBC to rethink ‘inappropriate’ reporting of UK economy • IPPR
The Institute for Public Policy Research (IPPR) is a centrist (in the British, not American, sense) thinktank on economics, and this is from its executive director Carys Roberts:
»
We are writing to you with regards to coverage of the 2020 Spending Review on Politics Live, Wednesday 25th November. Specifically, when responding to the Office for Budget Responsibility’s public sector borrowing projections, BBC News political editor Ms Kuenssberg said that “this is the credit card, the national mortgage, everything absolutely maxed out”, and later went on to comment that “for next few years, there is really no money”. We argue that this commentary misrepresents the financial constraints facing the UK government and reproduces a number of misconceptions surrounding macroeconomics and the public finances.
To focus on the “credit card” analogy, we would argue that this is never an appropriate metaphor for public finances. Maxing out a credit card would imply that the government is approaching a hard limit on its ability to borrow. This is not the case. It is the consensus amongst economists that the government should at this point in time not focus on reducing the deficit, but rather on delivering the spending necessary to secure a recovery from Covid-19. Modelling suggests that public debt as a proportion of GDP could actually fall were the government to embark upon a major investment package boosting jobs and growth, a position similar to that of the IMF in its flagship publication (pp 18-19) on the issue. This is in line with standard macroeconomic literature which stresses the beneficial effects of countercyclical government spending during crises.
Interest rates currently charged on government bonds are at record lows, so much so that the government is set to spend less on debt interest over the next five years than previously forecast, despite the rise in national debt over the course of the pandemic. Moreover, it is likely that interest rates will remain low for the foreseeable future; the interest rate charged on the 30-year gilt is currently 0.88%. These are not the signs of an institution approaching its credit limits.
«
Margaret Thatcher created the idea of the economy being like the household budget in the 1980s, and it stuck firmly in the national consciousness. It also happens to be wrong.
unique link to this extract
We talked to a PS5 scalper about how they got their consoles • Pocket Lint
Max Freeman-Wills:
»
After getting four consoles himself when pre-orders went live, each through manual means, Jack [the PS5 scalper] set about getting ready for launch-day stock. That meant, in practice, buying up the use of bots and proxies through connections on Discord, and choosing where they’d target come launch.
If that sounds hyper-technical, he says “it’s one of those things where once you get into it, it becomes very clear.” The Discord community helps newbies set stuff up, to avoid situations where someone “wastes a load of money on proxies” and badmouths the process. If people make money without too much hassle, more people will buy bots and the developers who act as the beating heart of reseller communities will continue making more and more money themselves.
Most communities can be found by carefully watching Twitter, he says -“you kind of have to gamble a bit on joining a group and finding one on Twitter, but I think Twitter always has been and always will be one of the best resources for releases like this because you’ve got your chronological timeline”. If you’re let in, the simplest step is to simply pay another reseller to run a bot for you – for a fee of around £50, you’ll have bought a far higher chance of getting a console than your average consumer, with a risk of detection and cancellation to go with it.
From there things have a steeper learning curve, but the picture Jack paints still summons some dismay – this isn’t something that’s hard to get into if you’re relatively comfortable using computers and browsing the web, which is hardly the highest bar to clear. That said, there is still an undeniable risk to be taken at some point when card details need to be handed over, and addresses (and mates’ addresses) detailed.
Those risks don’t disappear when you’ve got a console either – resellers regularly have their time wasted by frustrated non-buyers (something that’s easy to see the satisfaction in), which can be harmless but can make them feel vulnerable while it’s happening. In turn, he’s heard about a reseller who was apparently stabbed during a handover this week, showing that more unacceptable levels of anger are bubbling away at scalpers.
«
He targeted John Lewis, the big department store chain, on the basis that it probably wouldn’t be prepared for an onslaught. And picked correctly. Used to be that sites worried about bots taking them down; now it’s about cleaning them out. Except that’s good for John Lewis, and Sony.
unique link to this extract
How venture capitalists are deforming capitalism • The New Yorker
Charles Duhigg:
»
From the start, venture capitalists have presented their profession as an elevated calling. They weren’t mere speculators—they were midwives to innovation. The first V.C. firms were designed to make money by identifying and supporting the most brilliant startup ideas, providing the funds and the strategic advice that daring entrepreneurs needed in order to prosper. For decades, such boasts were merited. Genentech, which helped invent synthetic insulin, in the 1970s, succeeded in large part because of the stewardship of the venture capitalist Tom Perkins, whose company, Kleiner Perkins, made an initial hundred-thousand-dollar investment. Perkins demanded a seat on Genentech’s board of directors, and then began spending one afternoon a week in the startup’s offices, scrutinizing spending reports and browbeating inexperienced executives. In subsequent years, Kleiner Perkins nurtured such tech startups as Amazon, Google, Sun Microsystems, and Compaq. When Perkins died, in 2016, at the age of 84, an obituary in the Financial Times remembered him as “part of a new movement in finance that saw investors roll up their sleeves and play an active role in management.”
The V.C. industry has grown exponentially since Perkins’s heyday, but it has also become increasingly avaricious and cynical. It is now dominated by a few dozen firms, which, collectively, control hundreds of billions of dollars. Most professional V.C.s fit a narrow mold: according to surveys, just under half of them attended either Harvard or Stanford, and 80% are male. Although V.C.s depict themselves as perpetually on the hunt for radical business ideas, they often seem to be hyping the same Silicon Valley trends—and their managerial oversight has dwindled, making their investments look more like trading-floor bets. Steve Blank, an entrepreneur who currently teaches at Stanford’s engineering school, said, “I’ve watched the industry become a money-hungry mob. V.C.s today aren’t interested in the public good. They’re not interested in anything except optimizing their own profits and chasing the herd, and so they waste billions of dollars that could have gone to innovation that actually helps people.”
«
That’s the core of this piece, which is mostly about WeWork – which is of course the object lesson in VC gone wildly wrong. And what an object. (Theranos had almost no VC money.)
unique link to this extract
Astongate: fake emission figures, an embattled carmaker and a sock puppet PR company • LinkedIn
Michael Liebreich:
»
You may have seen the news story last week about a new report purporting to show that it takes 50,000 miles before an EV’s emissions beat those of a petrol car. First of all, the figures were comprehensively debunked – the correct figure is nearer 16,000 miles – then, over the weekend I uncovered evidence that the report was written by a sock-puppet PR company run from an address owned by Aston Martin’s Director of Global Government and Corporate Affairs.
Before I start, a caveat. This story is about getting the truth out EV carbon emissions, and how much lower they are than internal combustion cars. It does not deal with any other aspect of the environmental footprint of EVs, which are considerable and still require much investment and innovation. Nor does it show that EVs are superior to every other form of transport. Even the best EV will always have a carbon footprint, a material supply chain, and will cause particulate pollution. Active travel – walking, cycling, scooting and so on – should always be our first choice. With that said, let’s get stuck in!
“Astongate”, as I call it, started with widespread coverage of a new study, purporting to show that building an EV involves such huge CO2 emissions that you have to drive it 48,000 miles before it breaks even with a petrol equivalent. It was just the sort of story loved by UK media on the political right – always keen to pour cold water on any sign of green over-reach.
«
It’s quite the story, where an overzealous PR company with links to the car firms managed to plant a story which has been unravelling for days.
unique link to this extract
A lack of transparency is undermining pandemic policy • WIRED
Roxanne Khamsi:
»
I’ve seen this happen again and again since the start of the pandemic: a new, “science-based” Covid-19 measure is prescribed, but the science in support of it is either vague or missing altogether. Just last week, for example, I was working on a story about the latest research into quarantine procedures. The best data to this point suggests that an eight-day stretch of quarantine, combined with a Covid test, provides the same level of protection as the traditional 14-day quarantine. But then I saw New York state’s new policy: Some people who arrived from out of state are allowed to quarantine for just four days. I asked New York’s Department of Health how they’d come to this decision, and they sent me another statement from Cuomo, in which he said only that he’d “worked with global health experts” on the plan. A formal guidance from the state health department gave no research citations, either, but it did find space to boast about New York’s record of “strict adherence to data-driven, evidence-based protocols.”
This problem is hardly limited to one state. While reporting on that same quarantine story, I reached out to Alberta, Canada, which allows for an even riskier-seeming 48-hour period of quarantine for some travelers. What was the scientific basis for this policy? I never heard back.
A lack of transparency has even shown up in guidance from the World Health Organization. Back in March, I emailed the headquarters in Geneva to ask how they felt so certain at the time that the SARS-Cov-2 coronavirus was not “airborne.” The press office responded to my questions with a pair of unhelpful scientific documents. In that case, the decision to omit (or ignore) existing research—which suggested that other coronaviruses are likely to be spread by air—might well have been a deadly mistake.
«
It’s not so much a lack of transparency, as gigantic inertia – being both slow to start shifting from a position, and being slow to change course once headed in a direction. The WHO’s absurd refusal first to recommend masks and then to acknowledge that Covid transmits via aerosol has arguably cost a lot of time, and hence lives.
This is a consequence of organisations with strict organisational flows: they can’t flex. Social media flows more like water.
unique link to this extract
How an anti-lockdown ‘truthpaper’ bypasses online factcheckers • The Guardian
Jim Waterson:
»
photo that was spreading across Facebook, they were confused. The picture – which had been uploaded by users in the UK, US, Australia and elsewhere in the world – showed a headline that made the false claim that a US government agency had declared Covid-19 did not exist. It appeared to be from a real print newspaper, but no credible outlet would publish such a claim.
Then they had a breakthrough: it turned out the headline was from a new self-published conspiracy theorist “truthpaper” called the Light, edited by a man from Manchester who runs a business selling anti-vaccine T-shirts and 9/11 conspiracy merchandise.
The outlet, which has published three issues since it first appeared in September, draws heavily on the gloop of long-running online conspiracies about a new world order, which have attached themselves to the current pandemic. Among other things it encourages people to stop wearing masks and disobey lockdown on the basis that the coronavirus is a hoax.
It has been formed as a reaction to attempts by major tech platforms to clamp down on coronavirus disinformation – but the same tech outlets also help enable its reach: the Light’s distribution relies on a 5,000-strong private Facebook group where volunteers offer to hand out copies and post them through their neighbours’ doors.
«
Claims a circulation of 100,000 – which is sure to be untrue. That would be super-expensive. But shows how they’re trying to route around expectations by producing a printed thing: print is trusted.
unique link to this extract
Errata, corrigenda and ai no corrida: none notified
The Overspill: when there's more that I want to say 