Start Up: AlphaGo conquers chess, enter Amazon, Silicon Valley’s model problem, bitcoin’s future, and more

This will probably pay better than a Patreon account, data suggests. Photo by humbert15 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 12 links for you. Now with record approval ratings! I’m @charlesarthur on Twitter. Observations and links welcome.

Amazon wants a key to your house. I did it. I regretted it • The Washington Post

Geoffrey Fowler (and no, the boss – Bezos didn’t force him to do it or be nice about it):


The good news is nobody ran off with my boxes — or burgled my house.

The bad news is Amazon missed four of my in-home deliveries and charged me (on top of a Prime membership) for gear that occasionally jammed and makes it awkward to share my own door with people, apps, services — and, of course, retailers — other than Amazon.

“Amazon Key has had a positive reception from customers since its launch last month,” Amazon spokeswoman Kristen Kish said. “There have been situations where we haven’t gotten it right with a delivery and we use these situations to continue making improvements to the service.”

Big tech companies love building walled gardens, in ham-handed attempts to keep customers loyal. But for an ask this big (total access to your home, after all), Amazon needs to make Key better…

…When you use Amazon Key, you get a phone alert with a window when a delivery might occur. If no one is home, the delivery person taps an app that grants one-time access to unlock your door, places the package inside, then relocks the door. (They don’t recommend Key if you have a pet, and won’t come in if they hear barking.) The moment the door unlocks, the Cloud Cam starts recording — and sends you a live stream of the whole thing. It’s a surreal 15 seconds.


Not only but also: finicky setup, occasional bugs leading to fake warnings, and a door that ended up with Schrödinger’s Lock.
link to this extract

From territorial to functional sovereignty: the case of Amazon • Law and Political Economy

Frank Pasquale:


Economists tend to characterize the scope of regulation as a simple matter of expanding or contracting state power. But a political economy perspective emphasizes that social relations abhor a power vacuum. When state authority contracts, private parties fill the gap. That power can feel just as oppressive, and have effects just as pervasive, as garden variety administrative agency enforcement of civil law. As Robert Lee Hale stated, “There is government whenever one person or group can tell others what they must do and when those others have to obey or suffer a penalty.”

We are familiar with that power in employer-employee relationships, or when a massive firm extracts concessions from suppliers. But what about when a firm presumes to exercise juridical power, not as a party to a conflict, but the authority deciding it? I worry that such scenarios will become all the more common as massive digital platforms exercise more power over our commercial lives…

…For example: Who needs city housing regulators when AirBnB can use data-driven methods to effectively regulate room-letting, then house-letting, and eventually urban planning generally? Why not let Amazon have its own jurisdiction or charter city, or establish special judicial procedures for Foxconn? Some vanguardists of functional sovereignty believe online rating systems could replace state occupational licensure—so rather than having government boards credential workers, a platform like LinkedIn could collect star ratings on them.

In this and later posts, I want to explain how this shift from territorial to functional sovereignty is creating a new digital political economy. Amazon’s rise is instructive.


I was lucky enough to spend some time with Frank at Cambridge University earlier this year when he was a visiting fellow. He’s very incisive. His talk is here (on YouTube), if you have 16 minutes to spare. You do, right?
link to this extract

Three ways to remake the American economy for all • The Guardian

Senator Elizabeth Warren is a Democrat senator who might be a candidate for president in 2020. She gave a speech at the Open Markets Institute about dealing with monopoly power, especially in technology:


Donald Trump used to talk about the danger of monopoly. But that talk has pretty much disappeared now that he is president. Once he took the oath, he began stacking his administration with a who’s who of former lobbyists, Wall Street insiders, and corporate executives committed to tilting the scales even further in favor of their powerful friends and against everybody else. And just days ago, the Republican Congress handed out a giant tax giveaway to Wall Street corporations and the super-rich, leaving working families and college students to pick up the tab.

To rebuild an economy that works for everyone, not just the big guys, it is critical to reduce concentrated power in our markets. The federal government has the tools to do it; Congress handed antitrust enforcers those tools over a century ago. But those tools have been sitting on the shelf for decades, gathering dust.

Antitrust enforcers placed those tools on the shelf when they adopted Chicago School principles that narrowed the scope of antitrust laws; they moved away from the goal of protecting competition. It’s time to demand that antitrust enforcers pick up those tools, dust them off, and start enforcing the law again…

…It’s time to hold those corporations accountable for these competition-killing practices. And let’s be clear: holding everyone accountable means everyone. The investigation into Russia’s influence in the 2016 election has exposed how influential giant tech platforms can be. There is no exception in antitrust laws for big tech.

It’s time for antitrust enforcers to start looking critically at the ways in which massive amounts of data can be manipulated in ways that choke off competition.


link to this extract

Google’s AlphaZero destroys Stockfish in 100-game match •

Mike Klein:


Chess changed forever today. And maybe the rest of the world did, too.

A little more than a year after AlphaGo sensationally won against the top Go player, the artificial-intelligence program AlphaZero has obliterated the highest-rated chess engine. 

Stockfish, which for most top players is their go-to preparation tool, and which won the 2016 TCEC Championship and the 2017 Computer Chess Championship, didn’t stand a chance. AlphaZero won the closed-door, 100-game match with 28 wins, 72 draws, and zero losses.

Oh, and it took AlphaZero only four hours to “learn” chess. Sorry humans, you had a good run.

That’s right – the programmers of AlphaZero, housed within the DeepMind division of Google, had it use a type of “machine learning,” specifically reinforcement learning. Put more plainly, AlphaZero was not “taught” the game in the traditional sense. That means no opening book, no endgame tables, and apparently no complicated algorithms dissecting minute differences between center pawns and side pawns…

…GM Peter Heine Nielsen, the longtime second of World Champion GM Magnus Carlsen, is now on board with the FIDE president in one way: aliens. As he told, “After reading the paper but especially seeing the games I thought, well, I always wondered how it would be if a superior species landed on earth and showed us how they play chess. I feel now I know.”


The article includes one of the games. It feels quite different from how a human plays. AlphaGo seems to play as though it has all the time in the world; that it’s not particularly worried by threats, but equally wants to make exchanges on its own terms. Stockfish never seems to force it. AlphaZero even shows which openings are best. Queen’s Gambit and English Opening, apparently. (I prefer Bird’s Opening. Get things started.)

As Eric David notes at Silicon Angle:


What makes DeepMind’s latest accomplishment is noteworthy is the fact that it conquered three games with very different rule sets using a single AI. AlphaGo Zero, the latest version of AlphaGo, began “tabula rasa” without any prior knowledge or understanding of Go, shogi or chess, but the AI managed to achieve “superhuman performance” in all three games with stunning speed. IBM spent more than 10 years perfecting Deep Blue before it successfully mastered chess. AlphaGo Zero did it in just 24 hours.


link to this extract

Critical security flaws remain in smartwatches for kids • Forbrukerrådet

Norway’s Consumer Council on Mnemonic’s smartwatches, the “Gator” model:


Gator Norge gave the customers of the Gator2 watches a new Gator3 watch as compensation. The Gator3 watch turned out to have even more serious security flaws, storing parents and kids’ voice messages on an openly available webserver. The new watches also came with a significantly more expensive phone subscription.

In October, GPSforBarn launched the new app (GPSforalle) that works together with the watches. It contains similar security flaws as described with their previous app, the SeTracker. [in October 2017]

It is disconcerting that manufacturers, importers and retailers do not have better control over the products that they are selling. This is especially worrying when regarding safety-related products directed toward children, that could instead put the child in harm’s way, [Norway’s digital director of the Consumer Council] Finn Myrstad says.


The previous complaints were that “strangers can easily take control of the watch and track, eavesdrop on and communicate with the child. They may be able to track the child as it moves or make it look like the child is somewhere it is not. Some of the data is transmitted and stored without encryption.”

And they managed to make it worse?
link to this extract

After noon: digitalization wins the day – Nautilus Labs Logbook • Medium

Anthony DiMare on the strange way that ships report back to shore only at nautical noon:


On one particular voyage, a shipping company using Nautilus Platform noticed that the data collected directly from the ship’s engine showed a surprising variance in speed: the vessel ran at a higher speed during the day and at a lower speed overnight, while the average of the two was reported at noon.

If the shoreside teams had relied only on these reports, they would have misunderstood the vessel’s true performance. Total consumption would have been compared against the averaged speed — even though a ship requires exponentially more fuel to raise speed linearly.
In this case, the shoreside operator called the crew and inquired why they were seeing this behavior in the auto logged data. The operator had a simple request: please travel at a lower, consistent average speed for the rest of the journey. The net result was thousands of dollars of fuel saved — in that one leg alone of that one journey.

For most shipping companies, the prospect of saving a few thousand dollars on bunkers in one voyage isn’t that interesting. But it’s important to understand the long-term implications of this improved insight, as it impacts every decision our client would have made about the vessel in question. Let’s consider what would have happened, if the operator in this case didn’t have real-time visibility into the vessel’s actual performance.

If the crew continued to repeat the behavior without the operator’s knowledge, that vessel would have over-consumed hundreds of thousands of dollars worth of fuel over the course of a year — and millions of dollars in its practical lifespan.


So easy to forget how big savings can come from increased granularity in systems which have previously had the bare minimum.

link to this extract

Making Money from Data • DIGITS to DOLLARS

Jonathan Goldberg:


We recently spoke to the CTO of a large industrial company that manufactures big industrial systems. Like everyone else, they were trying to develop an IoT strategy. We sat with him while a software vendor was pitching their vision of the future, full of monetization possibilities. He was polite, at first, but after a while he broke in and said, “Before you go any further, you have to realize that almost all the data we capture is wort nothing.” That conversation only went down hill from there. However, he made a valid point. His company have been adding all kinds of sensors to their equipment for years.  They could capture petabytes a day, probably more. But 99.9% of that data essentially translated into “Status: Unchanged”.

We are not arguing that all data is worthless. However, we think it is clear that capturing value from data in the physical world is still a very poorly understood process. During the last Bubble in the 1990′,s we read a profile of a software company that had pitched its order system to a mid-sized produce distributor. After months of evaluation, the distributor determined that their existing fax-based system was still much more efficient than the fancy web-based system. It probably took another decade for software to bridge that gap. We think it may take that long for machine learning to make much difference to most companies.


Another decade of all the hype?
link to this extract

No one makes a living on Patreon • The Outline

Brent Knepper:


despite the revolutionary rhetoric, the success stories, and the goodwill that Patreon has generated, the numbers tell a different story.

Patreon now has 79,420 creators, according to Tom Boruta, a developer who tracks Patreon statistics under the name Graphtreon. (He has his own Patreon — “Graphtreon is creating Patreon graphs, statistics, and history” — which earns more than $500 a month.) Patreon lets creators hide the amount of money they are actually making, although the number of patrons is still public. Boruta’s numbers are based on the roughly 80% of creators who publicly share what they earn. Of those creators, only 1,393 — 2% — make the equivalent of federal minimum wage of $7.25 an hour, or $1,160 a month, in October 2017. Worse, if we change it to $15 per hour, a minimum wage slowly being adopted by states, that’s only 0.8% of all creators. In this small network designed to save struggling creatives, the money has still concentrated at the top.


This is the way of all networks, the way of the world: there are very few who are good at anything, and it’s always a pyramid. The only question is how wide the pyramid is; how sharp the slope from financial success to abject poverty.
link to this extract

Silicon Valley is sneaking models into this year’s holiday parties • Bloomberg

Sarah Frier:


Local modeling agencies, which work with Facebook- and Google-size companies as well as much smaller businesses and the occasional wealthy individual, say a record number of tech companies are quietly paying $50 to $200 an hour for each model hired solely to chat up attendees. For a typical party, scheduled for the weekend of Dec. 8, Cre8 Agency LLC is sending 25 women and 5 men, all good-looking, to hang out with “pretty much all men” who work for a large gaming company in San Francisco, says Cre8 President Farnaz Kermaani. The company, which she wouldn’t name, has handpicked the models based on photos, made them sign nondisclosure agreements, and given them names of employees to pretend they’re friends with, in case anyone asks why he’s never seen them around the foosball table.

“The companies don’t want their staff to be talking to someone and think, Oh, this person was hired to socialize with me,” says Kermaani, who’s sending models to seven tech parties in the same weekend.


Now they’re just going to suspect it of everyone, though.
link to this extract

CBOE to begin bitcoin futures trading December 10 • CoinDesk

Omkar Godbole:


The Chicago Board Options Exchange (CBOE) has announced that its planned bitcoin futures product will begin trading on Dec. 10.

In a statement published today, the firm said that trading would commence at 5 p.m. CT, with the first full day of trading starting that Monday. Trading on the CBOE Futures Exchange (CFE) under the “XBT” ticker, the company added in its release that trading of the futures product would be free through the end of December.

The announcement is a notable one given that a bitcoin future being launched by CME Group will go live the following week on Dec 18.

Ed Tilly, CBOE’s chairman and CEO, said in a statement: “Given the unprecedented interest in bitcoin, it’s vital we provide clients the trading tools to help them express their views and hedge their exposure. We are committed to encouraging fairness and liquidity in the bitcoin market. To promote this, we will initially offer XBT futures trading for free.”

The launch confirmation comes months after the Chicago-based exchange first detailed its plans to create a bitcoin futures product. At the time in August, the CBOE was working with New York-based bitcoin exchange Gemini, which is run by investors Cameron and Tyler Winklevoss, ahead of the launch.


Hmm. Can you have a working futures market in something that everyone – a phrase used loosely – seems to think will only increase in price? (Note I don’t say “value”.) Though it might create something of a brake if there’s enough money in the futures market betting on lower prices.
link to this extract

Bitcoin marketplace NiceHash gets hit by hackers who make off with millions in bitcoins • Mashable


NiceHash announced the thievery on their Facebook page, saying, “Clearly, this is a matter of deep concern and we are working hard to rectify the matter in the coming days.”

NiceHash’s head of marketing Andrej P. Škraba told the Wall Street Journal that an estimated 4,700 bitcoin were taken from the company’s bitcoin wallet.

One thing to note: as the value of a bitcoin continues to go up, so, too does the value of the heist. As of post time, the value of a single bitcoin has surpassed $15,000 (with prices on some local Korean exchanges already topping $19,000), meaning the value of the heist has, for the time being, surpassed $70m.


A couple of years ago it would have been $70,000. Timing is everything.
link to this extract

Don’t blame the US election on fake news. Blame it on the media • Columbia Journalism Review

Duncan Watts and David Rothschild:


While it may have been the case that the 20 most-shared fake news stories narrowly outperformed the 20 most-shared “real news” stories, the overall volume of stories produced by major newsrooms vastly outnumbers fake news. According to the same report, “The Washington Post produced more than 50,000 stories over the 18-month period, while The New York Times, CNN, and Huffington Post each published more than 30,000 stories.” Presumably not all of these stories were about the election, but each such story was also likely reported by many news outlets simultaneously. A rough estimate of thousands of election-related stories published by the mainstream media is therefore not unreasonable.

What did all these stories talk about? The research team investigated this question, counting sentences that appeared in mainstream media sources and classifying each as detailing one of several Clinton- or Trump-related issues. In particular, they classified each sentence as describing either a scandal (e.g., Clinton’s emails, Trump’s taxes) or a policy issue (Clinton and jobs, Trump and immigration). They found roughly four times as many Clinton-related sentences that described scandals as opposed to policies, whereas Trump-related sentences were one-and-a-half times as likely to be about policy as scandal. Given the sheer number of scandals in which Trump was implicated—sexual assault; the Trump Foundation; Trump University; redlining in his real-estate developments; insulting a Gold Star family; numerous instances of racist, misogynist, and otherwise offensive speech—it is striking that the media devoted more attention to his policies than to his personal failings. Even more striking, the various Clinton-related email scandals—her use of a private email server while secretary of state, as well as the DNC and John Podesta hacks—accounted for more sentences than all of Trump’s scandals combined (65,000 vs. 40,000) and more than twice as many as were devoted to all of her policy positions.

To reiterate, these 65,000 sentences were written not by Russian hackers, but overwhelmingly by professional journalists employed at mainstream news organizations, such as The New York Times, The Washington Post, and The Wall Street Journal. In just six days, The New York Times ran as many cover stories about Hillary Clinton’s emails as they did about all policy issues combined in the 69 days leading up to the election.


link to this extract

Errata, corrigenda and ai no corrida: none notified

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.