Sunday, March 27, 2016

Stuff I'm reading

One thing about a rainy first half of March: my allergies this weekend are Out Of Control!

  • Unsigned Deals and Roguish Bonds

    Bloomberg's Matt Levine might be the best writer you haven't heard of, and Friday's column was Levine at his best, as he worked to illuminate the arcane corners of modern Financial Engineering with his characteristic tongue-firmly-in-cheek style.

    Among various other topics, he digs into the Credit Suisse news:

    What is most conceptually wonderful about this trade is that Credit Suisse is selling investors the risk that Credit Suisse is committing fraud. But what if Credit Suisse is committing fraud when it sells the bonds? Can the bondholders sue Credit Suisse for fraud? If they win, do they have to pay Credit Suisse back? It is a Möbius strip of liability, a Klein bottle of derivatives, a mathematical object that exists solely to prove that any formally complete system of capital regulation is inconsistent, or at least weird.
    Levine, of course, has been writing wonderful material about the financial world for years, and Friday's article references two other wonderful Levine essays:

    • Marvel At The Derivative On Its Derivatives That Credit Suisse Wrote To Itself
      Credit Suisse has sold CDS to itself on its derivatives counterparties – if any money changes hands it will go from CS, to the CDS seller, back to CS, which may explain why it was able to find someone to write such a weird CDS contract. If CS’s other counterparties’ spreads widen then Credit Suisse will have an offsetting gain on the CDS that it bought, but because of the accounting differences it will not have an offsetting, offsetting loss on the CDS (sorry, “credit support facility”) that it sold. It has transformed a mark-to-market CVA exposure that would reduce its net income (and, thus, regulatory capital) if counterparty spreads widen, into an accrual thing that won’t reduce capital until there are actual defaults on CS’s actual derivatives.
    • Mourn For The Derivative On Its Derivatives That Credit Suisse Wrote To Itself
      The capital regulators decided that their trade was not an economic contract, so it won’t have the capital effect they wanted, so it would appear to be back to the drawing board.

      Delightfully they do seem to be going back to the drawing board, rather than conceding defeat; IFR quotes someone in the bank as saying that “continued evolution of the Basel III rules is likely to require some modifications to the hedge. These changes are already well advanced.” This may work out yet. My faith in Credit Suisse is diminished, but not destroyed. You can come back from this, Credit Suisse. You have to. For all of us.

Meanwhile, in other things to read...
  • From QA to Engineering Productivity
    SETs initially focused on building tools for reducing the testing cycle time, since that was the most manually intensive and time consuming phase of getting product code into production. We made some of these tools available to the software development community: webdriver improvements, protractor, espresso, EarlGrey, martian proxy, karma, and GoogleTest. SETs were interested in sharing and collaborating with others in the industry and established conferences. The industry has also embraced the Test Engineering discipline, as other companies hired software engineers into similar roles, published articles, and drove Test-Driven Development into mainstream practices.
  • Google DeepMind's AlphaGo: How it works
    Monte Carlo Tree Search (MCTS) is an alternative approach to searching the game tree. The idea is to run many game simulations. Each simulation starts at the current game state and stops when the game is won by one of the two players. At first, the simulations are completely random: actions are chosen randomly at each state, for both players. At each simulation, some values are stored, such as how often each node has been visited, and how often this has led to a win. These numbers guide the later simulations in selecting actions (simulations thus become less and less random). The more simulations are executed, the more accurate these numbers become at selecting winning moves. It can be shown that as the number of simulations grows, MCTS indeed converges to optimal play.
  • GDC2016 FlashBackward
    MMOs gave you game guilds. They gave you free to play. They gave you the profession now called community manager. They birthed the farming game that became social gaming. Would there be bitcoin today without the paths first explored by gold sellers? There certainly wouldn’t be a Minecraft without MUD.

    Aspects of MMOs gave you Facebook itself: a world with no world there, one that maybe hasn’t listened closely enough to the old lessons on player rights and governance. What is Twitter but public chat for the world? What is Facebook, LinkedIn, but your character sheet, reinforcing rather than challenging our notions of identity? We are all avatars now, but perhaps our worlds are a little more mundane.

  • Amidst of the Rubble of Bedrock City: The rise and fall of the Flintstone empire
    During a visit last summer, both parks were more like ghost cities. In what was supposed to be the high season for both parks, you could have Bedrock City almost to yourself. In South Dakota, a small family explored the fake town, while another took a ride in Fred’s car, famously operated by foot. (In this case, it was operated by a small motor.) If you wanted, you could poke your head into the oval openings of a custom cutout board and take pictures pretending to be the freakishly strong Bamm-Bamm, holding a massive stone barbell over his head using only the tip of his index finger.


    In Arizona, the sound of The Flintstones theme song blared across the flat expanse of park from the patchy sound system of the movie theater in which no one was watching old Flintstones episodes. No one waited in line for the brontosaurus slide either. A few handfuls of adults roamed from house to house taking selfies, petting the goat inexplicably penned up near the entrance to the park, or sitting in the small but pleasant garden that Linda planted in the nineties, after her husband had passed.

  • Shuffling Michigan
    Their approach is to resize each county by population (# of total votes would be good, but no good could come of remaking these on the fly as votes come in), and try to keep the overall shape of the state. Unfortunately, in so doing, they shuffle the counties around any old which way. The Lower Peninsula of Michigan has 68 counties in reality, the Upper Peninsula has 15. But Decision Desk HQ has shoved most of the counties into the Upper Peninsula, which now has 58, vs. 25 that remain in the Lower Peninsula.

    This means that we can’t really see spatial patterns, which is sort of the point of having a map. Notice how, on the NYT map, the gold-colored counties that went for Cruz are mostly clustered in the west. The Decision Desk HQ map keeps some of them there (in yellow), but scatters the rest of that cluster around the Upper Peninsula. The two purple counties both went for Kasich, and Decision Desk HQ would have us think they are neighbors (which would be an interesting thing to note as far as spatial patterns go), but in fact they’re the two teal-colored counties on the NYT map that are on opposite sides of the state.

  • Why does the United States keep killing #2 in ISIS?
    As Zack mentions, there may be reasons why the #1 is harder to find and kill, but I would suggest a complementary hypothesis. At many points in time there is more than one #2, just as corporations may have a variety of Executive Vice Presidents.

    If you a leader of a terror group, do you really want a well-defined #2 who is a focal alternative and who can move to overthrow you? Or do you prefer seven competing #2s, with somewhat unclear status, whom you can play off against each other, or make compete against each other, and offer various sticks and carrots and promotions of influence against each other?

    And let’s say that one of these #2’s is killed. How will the United States report this? “One of seven #2’s has been killed”? Or perhaps the easier to communicate and more important sounding “We have taken out number two.”

  • National Hero
    There is nothing glorious about any of it. People don’t die gloriously for their beliefs. They die instantly or silently or crying out in pain.

    The notion of tactically risible but symbolically meaningful blood sacrifice is one that angry and stymied young men have always embraced, not least this week in Brussels. There is nothing new about disenfranchised twenty-somethings appropriating the images and ideas of whatever religion they happen to grow up around to tart up the essentially adolescent idea that blood cleanses, especially the blood of others.

    What we now call radicalisation is simply the age-old desire of the young to believe in purity; to believe in it so completely that it comes above human life. But purity does not exist. Humanity isn’t good enough at any single thing to make it more important than the irreplaceable consciousness of just one of us.

No comments:

Post a Comment