(Bloomberg View) -- Money laundering.
In 2012, Citigroup Inc. got in trouble for not having good enough systems to prevent and report money laundering. The Office of the Comptroller of the Currency, one of Citi's regulators, ordered the bank to get its act together, in numbing detail. The 2012 consent order goes on for 22 pages and covers things like setting up a compliance committee, clarifying lines of responsibility for managing anti-money-laundering compliance, adopting new customer due diligence programs and suspicious-activity-reporting policies, and hiring a consultant to review past activity for reporting failures. Each item has a pretty comprehensive list of requirements for how Citi was supposed to implement it in order to satisfy the regulator.
And then Citi ... didn't do it? So the OCC fined it $70 million "for failing to comply with the agency's 2012 consent order." It's not exactly clear what Citi didn't do -- the OCC "didn't specify its complaints" -- though for $70 million you'd hope there were a lot of failures. You'd hope that they forgot about the consent order entirely, just put it in a drawer somewhere, and then five years later were like "oh wait we were supposed to do something about that?"
It's really weird. In general, I have quite a bit of sympathy with banks who get into regulatory trouble for stuff like not having good enough anti-money-laundering controls. Banks are big, and the rules are complicated, and you'll never set up a perfect system to catch money laundering, and if any money laundering does slip through the cracks then you will look bad in hindsight. From the perspective of a bank executive, compliance is a statistical process: You can never achieve perfect compliance, you can never have zero illegal activity, you have to manage your compliance program to some tolerable level of illegality, and no one will ever tell you what that tolerable level is. (No regulator will say "eh, if you launder money five times a quarter, that's fine.") It is difficult and full of uncertainty and even your best efforts may not be good enough.
But this is, like, they had a checklist! Since the financial crisis banks have spent limitless amounts of money on compliance; surely Citi could have hired some people to go through the OCC's list and actually do the stuff. The hard job in compliance is looking ahead, figuring out what is going on that will eventually get you in trouble and nipping it in the bud even before the regulators become interested in it. This, by contrast, was the lowest-hanging fruit: Citi had already gotten in trouble, and the OCC gave it an explicit formula for how to fix it, and Citi still messed it up.
Deep learning.
"Controlling the physical world is an integral part and arguably a prerequisite of general intelligence," begins this paper from Alphabet Inc.'s DeepMind team, which has used reinforcement learning techniques to teach computers to play chess and Go with a level of skill and insight beyond that of any human player. The computers learned through "self-play" in which they started out knowing only the rules of the game; they had no built-in human preconceptions about how to play, and were able to find approaches that humans never would have.
Now DeepMind wants to do the same thing for "controlling the physical world," which, you know, I am not sure I want robots with no preconceived human notions to suddenly become better at "controlling the physical world" than humans are, but at least they are starting small. For one thing, they are starting with a physics simulator rather than the actual physical world; for another thing, the tasks that the computer is learning how to do are just standing, walking and running.
Obviously humans have been standing, walking and running for many thousands of years, so you'd think that we would have developed some expertise. But DeepMind's computers quickly found solutions to those problems that, I think it is fair to say, humans would not have. Here is a video from DeepMind showing those solutions, and what you're going to want to focus on is the part from 1:39 to 2:15 in which the simulated humanoid stands, walks and runs. This video, I think it is fair to say, is good. Here, for instance, is how the deep-learning neural network solved the problem of standing:
Not bad! It's a convincing stand. It is not I think what most humans would do with their elbows, but that's what you get when you eliminate preconceived notions and just let the computer figure out the problem for itself.
Here's one solution to the problem of running:
Again DeepMind has made some real innovations in the use of elbows but, yes, that looks like running. Here, though, is another solution:
That ... yes. That's it. That's how you run now.
I am sure that this research has important implications for the use of deep learning techniques in investing but for now let's just enjoy watching that little green robot run.
Uber.
There's nothing all that unusual about the fact that Travis Kalanick co-founded Uber Technologies Inc., built it into a multibillion-dollar company, watched it go public, cashed out some of his founder's shares to become a billionaire, and then was forced out as chief executive officer by activist shareholders who concluded that for all his visionary abilities he was not suited to be the CEO of a public company. What is weird is the order of those events: Kalanick was forced out last year, he'll cash out $1.4 billion worth of stock later this month, and Uber is aiming to go public next year. Usually you have to go public to cash out that much, and you have to be public before the activists come for you. But in my model Uber is a large public company that happens to be private, and that keeps being relevant. Anyway:
Former Uber Technologies Inc. Chief Executive Officer Travis Kalanick, who has long boasted that he's never sold any shares in the company he co-founded, plans to sell about 29 percent of his stake in the ride-hailing company, people with knowledge of the matter said.
Kalanick stands to reap about $1.4 billion from the transaction with SoftBank Group Corp. and a consortium of investors who have agreed to buy equity valuing Uber at $48 billion, said the people, who asked not to be identified discussing private negotiations.
"Why would you ever do an initial public offering," is the question I always find myself asking when Uber does things like this. If you can found a company, scale it, leave your job and cash out a billion dollars, then that is pretty much the whole life cycle. If you can do all that without going public or being acquired, who needs public markets?
Intel.
Speaking of CEOs who cash out, it is perhaps a little awkward that Intel Corp. CEO Brian Krzanich "sold off a chunk of company stock in the fourth quarter of last year, bringing his holdings to a five-year low," months after Intel had been alerted to a security vulnerability in its chips, but also a month before news of that vulnerability became public. I am always skeptical that stories like this are actually about insider trading, just because it would be too dumb and obvious for Krzanich to dump his stock just before announcing bad news. Also: The news doesn't seem to have been that bad for Intel's stock, and in fact Krzanich's sales last year were mostly at prices below yesterday's close. As a diabolical plan it seems pretty unimpressive.
And in fact Intel has an explanation:
“Brian's sale is unrelated,” a company spokesman said. “It was made pursuant to a pre-arranged stock sale plan with an automated sale schedule. He continues to hold shares in line with corporate guidelines.”
Krzanich's plan seems to involve getting stock grants at the beginning of each year and then selling as much as he can in the fourth quarter, which he has done consistently for a few years. And he does "hold shares in line with corporate guidelines," in the sense that Intel requires its officers to hold a certain number of shares, and his sales in November brought him down to the exact minimum requirement. I guess that is fine: The point of the minimum requirement is that you have to hold that many shares, and he does. Still it seems like there might be a shadow requirement that the CEO should own way more than the minimum in order to demonstrate enthusiasm. Certainly if I worked at a public company and got much of my compensation in stock, I'd sell as much as I could as soon as I could, but that -- I always assumed -- is part of why I'll never be a public company CEO. Perhaps I have misunderstood the expectations, though.
Blockchain blockchain blockchain.
Generally speaking I am pretty skeptical of stunt-y applications of "the blockchain" in which a single central intermediary keeps a database of things and then decides one day to call that database a "blockchain." But here is Sarah Holder with a pretty interesting one:
In 2014, Moscow Mayor Sergey Sobyanin launched Active Citizen, an e-voting platform designed to allow citizens to directly weigh in on non-political city decisions—things like setting speed limits, plotting bus routes, and naming subway stations. Since then, 2,800 polls have been administered via the app and almost 2 million users across this city of 11 million residents have participated.
But now it will be replacing the word "app," which I guess was pretty modern in 2014, with the word "blockchain":
“We are excited to improve [the] credibility and transparency of e-voting system in Moscow by introducing blockchain,” said Artem Ermolaev, the city's chief information officer, in a statement. “We believe that blockchain will increase trust between the citizens and the government.” ...
The publicly-owned nodes have full access to a personal copy of the database even before results are published city-wide, and are able to cross-reference every block with the official counts. This, the city insists, will give citizens incorruptible evidence of how many votes were cast, when, and for what. “[Each node] stores the copy of votes database and will be notified if one block has been changed or deleted,” said a spokesperson for the Moscow Smart City Lab, in an email. “Thus, there is no possibility of rigging the vote.”
The basic use case of a blockchain is for applications where you don't trust a single central administrator, so you build a distributed database that is maintained by all the users of the system. By that standard, few blockchain projects announced by banks and companies and governments make any sense, because so many of their blockchains are maintained by a central administrator whom you need to trust to bother using the blockchain thing in the first place. If, say, a gold exchange keeps a record of who owns its gold "on the blockchain," that doesn't really solve any trust problems, because if you want to actually get the gold you have to get it from the exchange, so you have to trust the exchange. If the exchange steals the gold, then your blockchain entitlement doesn't do you any good.
But Moscow's experiment suggests a different use case: This is a blockchain for a centralized database where you don't trust the central administrator but have to leave decisions to it anyway. Putting Moscow's bus routes "on the blockchain" won't prevent Moscow's government from setting up the bus routes however it wants. But putting the polls "on the blockchain" will at least let people see what the vote was: The government can ignore the non-binding vote, but if it does then you'll know about it. The blockchain gives you information about how much, or how little, you should trust the central intermediary. It only gets you transparency, not power, but still that is something.
Crypto crypto crypto.
Here's a story about how a guy who co-founded Ripple was the fifth-richest person in the world for a while yesterday, passing Mark Zuckerberg, due to the soaring value of Ripple's XRP tokens. "Well but try to spend it" is one possible response, and while that is not wholly satisfying -- Zuckerberg might have trouble spending billions of dollars' worth of Facebook stock all at once -- there is probably something to it. "This is beyond insane," and "everyone in the industry is now slinging crack crypto cocaine to retail addicts," are among the representative quotes in the story. Also: "Ripple has so far announced that one company, a Mexican money-transfer business, is planning to use the Ripple token." That's the kind of value-add that creates the fifth-largest fortune in the world these days. As long as it's on the blockchain.
On the other hand: "Ripple Slides After Coinbase Says Not Adding New Crypto Coins." And here's yet another argument that bitcoin is bad because no one wants to spend it. And here is "Startup Turns a Buzzword Triple Play With Fintech, Bitcoin, Weed," though the timing on that one is a little rough: Here's "Justice Department's Marijuana Reversal Puts Damper on Hot Stocks." It's okay though: Even if you can't run a legal weed business any more, you can still run a weed business on the blockchain. Everything's fine on the blockchain. This is of course not legal advice, not even on the blockchain.
Happy Dow 25,000!
Yesterday morning the Dow Jones Industrial Average crossed 25,000 for the first time, while bitcoin languished in the $15,000s. Now I will have to throw out all my "Bitcoin 25,000 Before Dow 25,000" hats, which I ordered back when the Dow was adding 1,000 points every two months and bitcoin was adding $1,000 every few days. One lesson here is that nothing that you read in this column is ever investing advice. Another lesson is, I don't know, maybe bitcoin has become a stable and mature store of value instead of just a rocket to whatever arbitrary large number you can think of? Anyway, if you want any of the hats -- now a rare collector's item -- before I throw them out, they are yours for the now relatively low cost of one bitcoin each.
Things happen.
Global Debt Hits Record $233 Trillion. Fifty shades of green bank capital. Investors pour money into funds that protect against inflation. An MBA is still a great boost for salaries. How Spotify solved a $1 billion debt problem that will help it IPO. Mark Zuckerberg's New Year's Resolution: ‘Fixing' Facebook. Subaru's Plan to Woo Americans: A Roomy SUV With 19 Cup Holders. Next Up for the Northeast: Howling Winds and Frigid Air. It's so cold in the US that sharks are freezing to death.
If you'd like to get Money Stuff in handy email form, right in your inbox, please subscribe at this link. Thanks!
This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
Matt Levine is a Bloomberg View columnist. He was an editor of Dealbreaker, an investment banker at Goldman Sachs, a mergers and acquisitions lawyer at Wachtell, Lipton, Rosen & Katz and a clerk for the U.S. Court of Appeals for the Third Circuit.
To contact the editor responsible for this story: James Greiff at jgreiff@bloomberg.net.
For more columns from Bloomberg View, visit http://www.bloomberg.com/view.
©2018 Bloomberg L.P.
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.
