Icosian Reflections

…a tendency to systematize and a keen sense

that we live in a broken world.

December in Review: Part I

Previously: December in Review, including what we're doing here.

Part I (this part) covers, inter alia, mathy stuff (quantum computing, math proper, AI theory), computer security, free speech law, liberalism, economics, and Wall Street. Part II will cover medicine, lifehacks, effective altruism, and collected miscellany.

A brief meta-index: (skip to first)

  • Shtetl-Optimized, Scott Aaronson breaking down theoretical quantum computing into terms I can (usualy) understand: (SO.1) Did Google really unveil a quantum computer that's a hundred million times faster? (not really.) (SO.2) 10,000-dimensional physics is uncomputable. (really.)
  • What's New, Terence Tao doing math that makes my eyes glaze over: (WN.1) "A conjectural local Fourier-uniformity of the Liouville function". (WN.2) Decoupling, i.e. "almost orthogonality", and a few theorems. (WN.3) A Polymath proposal for explaining two remarkable identities in irreducible polynomials.
  • Still Thinking, Allen Yuan just getting started on a topology-focused math blog by explaining cohomology theory(ies?) (ST.1) Intro; (ST.2) Cohomology 101.
  • A.Critch Blog, Andrew Critch writing about AI decision theory and other germane curiosities: (AC.1-4) A brainteaser and mathematical analysis of a counterintuitive result about a Prisoners'-Dilemma-playing AI. (AC.5) On the importance of advocating for AI safety. (AC.6) "Why I want humanity to survive."
  • Schneier on Security, Bruce Schneier explaining the things we should and should not be worried about w/r/t privacy and computer security: (SOS.1) "The Moral Character of Cryptographic Work". (SOS.2-4) Headlines on Israel, intelligent cars, and burglary footage
  • Popehat, Ken White and Mark Randazza providing utterly irreverent legal analysis and/or social commentary: (PH.1) Planned Parenthood threats, (PH.2) cybersquatting on jebbush.com, (PH.3) regulating doctors, (PH.4) arresting jury-educators, (PH.5) personal attacks in the media, (PH.6-8) free speech is worth defending (PH.9) Star Wars [no spoilers] (PH.10) "Talking Productively About Guns".
  • Interfaces of the Word, Freddie de Boer on...stuff. (IW.1-2) Teaching English like it's important. (IW.3) The first counterpoint to the "are nerds oppressed" yelling-match that's made me sit up and think. (IW.4-6) Seeking liberal reform like we mean it.
  • Marginal Revolution, Tyler Cowen and Alex Tabarrok, professors of economics at GMU, cover pretty much every event of economic import in the world. (MR.1-14) Here are just some of their best posts in December.
  • Vox: Dylan Matthews, who tends to focus on economic, altruistic, and/or other social issues. (DM.1) Finland's basic income experiment, explained; (DM.2) fighting poverty by reminding people to claim their EITC is surprisingly...effective?
  • Money Stuff, Matt Levine covers the goings-on of Wall Street with a sense of humor that will make the most humorless banker giggle. (ML.1) The best press release about a truck heist you'll ever read. (ML.2) The metaphysics of the Federal Reserve rate. (ML.3) The story of how Yahoo!, Inc. is worth negative thirteen billion dollars. (ML.4) Twin traders run a scam by switching trading accounts. (ML.5) Why making the exchanges slower isn't an easy solution to high-frequency trading. (ML.6-8) Hillary Clinton and Bernie Sanders's plans to "rein in Wall Street", compared with some commentary.
###(Shetl-Optimized) ####[Scott Aaronson](http://www.scottaaronson.com/blog/)

Once upon a time, I was introduced to Scott Aaronson's blog, which is incredibly well-written and eminently readable when it addresses topics technically accessible to mere mortals; incidentally, Aaronson's clarity helps to expand that category to include basic theory of quantum computation.

When I later had the fortune to take a course with him, I found him to be one of the best professors I've had in college -- and now, I can actually understand posts like Shtetl-Optimized | Google, D-Wave, and the case of the factor-10^8 speedup (for what?).

Basically, as Aaronson reports: D-Wave -- the quantum computing company now owned by Google and founded on the premise that quantity is more important than either quality or theoretical rigor -- announced that their new quantum computer, the 2X, runs a hundred million times faster than...a simulated version of the 2X running on a classical computer. For the trigger-happy pop-sci news media, this is enough to be off to the races, but Aaronson actually bothers to sift through the science:

Yes, there’s a factor-108 speedup that looks clearly asymptotic in nature, and there’s also a factor-108 speedup over Quantum Monte Carlo. [Note that Quantum Monte Carlo, despite its name, is a classical algorithm for simulating quantum phenomena. -ry] But the asymptotic speedup is only if you compare against simulated annealing, while the speedup over Quantum Monte Carlo is only constant-factor, not asymptotic.

And in any case, both speedups disappear if you compare against other classical algorithms, like that of Alex Selby. Also, the constant-factor speedup probably has less to do with quantum mechanics than with the fact that D-Wave built extremely specialized hardware, which was then compared against a classical chip on the problem of simulating the specialized hardware itself (i.e., on Ising spin minimization instances with the topology of D-Wave’s Chimera graph).

Thus, while there’s been genuine, interesting progress, it remains uncertain whether D-Wave’s approach will lead to speedups over the best known classical algorithms, let alone to speedups over the best known classical algorithms that are also asymptotic or also of practical importance. Indeed, all of these points also remain uncertain for quantum annealing as a whole.

Note that quantum annealing, the method that D-Wave uses, is a nonstandard computing model which is likely not the same thing as general gate-based quantum logic (the natural extension of Turing's computing model).

Further reading: see also Google's original paper, an explanatory summary by Zintchenko, Brown, and Troyer putting the whole thing in context, and Scott A.'s Q&A with MIT News.

Of course, Scott A.'s real academic hobby is asking the question That's very nice in theory, but what does it tell us, philosophically, about physics?

He's absolutely in his element in Shtetl-Optimized | Ask an unbounded question, get an uncomputable answer. Here, he tackles a 146-page paper recently published in Nature by Cubitt, Perez-Garcia, and Wolf which proves an absolutely spectacular result about the computability of 10,000-dimensional physics.

###(What's New) ####[Terence Tao](https://terrytao.wordpress.com/)

Iff you wanted to just read math, maybe try Terry Tao's blog? I admit that I can't get past the first statement of the main result on most days, but I keep scanning post titles in the idle hope that he gives me something that's useful for proving results in the tameness of algebraic amplitudes, a question near to my heart as it was the first serious mathematical research question I ever asked. (Incidentally, I discovered and wrote about it in Scott Aaronson's course on Quantum Complexity Thoery).

What's New | A conjectural local Fourier-uniformity of the Liouville function made me do a double-take -- my paper explores the connections between generalized Liouville numbers and tameness -- but unfortunately, Joe Liouville was one of those people with so many things named after them that the Liouville function has nothing to do with Liouville numbers as far as I can tell.

Meanwhile, What's New | Decoupling and the Bourgain-Demeter-Guth proof of the Vinogradov main conjecture was fascinating reading until I gave up, and its sequel is probably good as well. Also I have no idea what's going on in Terry's Explaining identities for irreducible polynomials (on the Polymath group blog) except that now that he's pointed it out, I really want some kind of an answer.

###(Still Thinking) ####[Allen Yuan](https://allenyuanblog.wordpress.com/)

Iff you still want math reading, my friend and classmate Allen Yuan has started a blog on topology and cohomology. It's also above my pay grade (or at least, the pay grade of me without pen and paper and time to take it seriously), but it's clear and well-written and probably the sort of thing I'd be interested in if topology were my cup of tea. (It's not.)

###(A.Critch Blog) ####[Andrew Critch](http://acritch.com/blog/)

Critch is an awesome-cool guy to work with, and while I was sad to hear that he's no longer working at my future employer, I am happy to hear that he's decided to work on the hard problem of making sure future artificial intelligences don't accidentally kill us.

If you don't think that this is an especially important problem, because, well, we're making the AIs, then buckle in for this brainteaser: A.Critch | The Problem of Indignation Bot, Part 1. Thesis: "This very simple AI for playing the Prisoner's Dilemma sometimes does exactly the opposite of what you'd expect. Here's why. Here's more math on how to make it work." Before reading through this brainteaser/blog-post-sequence, I thought I had a decent idea of what sorts of research the Machine Intelligence Research Initiative did, but now I think I grok it at a deeper level.

More on MIRI, Critch writes about why A.Critch | MIRI needs funding to scale with other AI safety programs, as well as A.Critch | Why I want humanity to survive, which is just beautiful and excellent.

Incidentally, Luke Muehlhauser's If you're an "AI safety lurker," now would be a good time to de-lurk is timely and relevant.

###(Schneier on Security) ####[Bruce Schneier](https://www.schneier.com/)

Bruce Schneier (fellow at Harvard's Berkman Center for Internet and Society, board member at the Electronic Frontier Foundaction, &c. &c.) concerns himself primarily with a more mundane danger: the twin issues of computer security and government surveillance.

First, h/t Bruce, go read this paper by Phillip Rogaway titled The Moral Character of Cryptographic Work. Rogaway harkens to the Russell--Einstein manifesto, but in my opinion, has penned a document which is much its modern counterpart. On the object level, he's writing to cryptographers, but I do believe that scientists in every field should read it and ask: Where is laid out the moral call of my field? And if it has not yet been written, consider writing it yourself. (Also, read the Russell--Einstein Manifesto, because damn, people used to write prose.)

The proper Schneier on Security I've got all have self-recommending titles:

A propos of nothing, a Kickstartup is selling tinfoil signal-resistant hats.

Somebody might say that it's a specially designed hat for bouncing electromagnetic waves and radiation. But is it really so? Yes, it reflects signals from cell phones, wi-fi routers, microwaves and it generally blocks waves transmitted from electric devices. It's not blocking for 100% percent, but better than nothing. Moreover, it looks really cool! It is one of the most comfortable and functional headwear you have ever worn. (...)

In the words of my generation, I just can't even.

###(Popehat) ####[Ken White, Marc Randazza, various](https://popehat.com/)

One of the good things about reading Schneier on Security is that I'm pretty confident that, if anything truly important happens in the world of security, I will probably hear about it (and, from a history of reading, actually understand the context). Popehat is like that, but for freedom-of-speech law. And also usually funnier.

Sometimes it's legal analysis with self-recommending titles:

Sometimes it's a bit more editorial (and, in my opinion, more fun):

Sometimes it's vicious:

aside: Posner's showed up here on Faults before -- or more properly, in a piece I wrote for the Harvard Political Review -- in a counter-opinion to a (bad) piece he wrote for Slate on effective altruism.

Sometimes it's about the magic of childhood and Star Wars:

And sometimes it's really, incredibly important:

###(Interfaces of the Word) ####[Freddie de Boer](http://fredrikdeboer.com/blog/)

Well, if we're already in the contrarian-left corner of the Internet...it shouldn't come as much of a surprise that I really like Freddie. He gives me hope that 'professor of English' can, if done right, be an incredibly important job. He takes critique-as-reflective-of-culture seriously (Interfaces of the Word | this modern anxiety); he takes his academic role as a teacher with deadly gravity (Interfaces of the Word | Teaching Style to Freshmen: Why & How); but of course...those aren't really what he's Internet-famous for. He

For a year which was ushered in by Scott Aaronson's "Comment 171" and the ensuing acrimonious debate over whether or not it is possible as a matter of social fact for feminists to oppress nerds, I think that Interfaces of the Word | that's not why you feel the way you feel is the appropriate closing bookend. It's the first post that I've read -- in the entire past year -- that honestly acknowledges the pain and rejection that the other side feels...and applies that understanding to producing answers to the questions So what? What do we do now? I don't agree entirely, or even 95% with Freddie (I don't think he gives enough credence to possibility of oppression within privilege.), but at least I can see it now -- the things that "my side" might be missing, and the introspection I need to do in order to figure out where I'm coming from in all this.

I don't agree with him on all things, but I do agree with two of his major themata: that the liberal movement is far weaker than most in the leftish echo chamber realize, and that real liberal reform would look very little like the contemporary social justice movement does. In his words:

The last of these, I believe, seriously ought to be required reading for anyone hoping to make the world better by liberal reform.

###(Marginal Revolution) ####[Tyler Cowen, Alex Tabarrok](http://marginalrevolution.com/)

Something tells me that Freddie wouldn't get along so well with right-leaning economist Tyler Cowen, but at the end of the day, de Boer's focus on material liberalism has me thinking: leftists should read more economists if they're serious about (1) making the world better and (2) winning. For example: Marginal Revolution | Paul Krugman reviews Robert Reich -- which asks, does income inequality come from monopoly power? I think anyone who cares about income inequality might care about that one...

For that matter, is it a product of assortative mating? (Probably not; Marginal Revolution | How much does assortative mating matter for income inequality?)

Other economic questions:

Tyler is also famous (notorious?) for daily linkwraps, pointing to things like:

And both Tyler and Alex have a knack for pulling quotes and recommendations from an utterly inhuman quantity of reading:

The blog is generally light on sit-there-while-I-talk-at-you material, but when it's there, it's good:

If you want ten more posts, Marginal Revolution | The Top Ten MR Posts of 2015. I remember reading most of them through the year, but they were worth re-reading anyway.

###(Vox: Dylan Matthews) ####[Dylan Matthews](http://www.vox.com/authors/dylan)

Elsewhere in reporting on economic happenings-in-the-world: I don't read all of Vox, but I do generally find Dylan Matthews's posts to be good reading. The titles mostly speak for themselves, with the angle, spin, and slant exactly as they appear on the tin:

###(Money Stuff) ####[Matt Levine](http://www.bloombergview.com/contributors/matt-levine)

If you want to read about Wall Street goings-on and come away from the experience taking the whole financial sector exactly zero percent seriously, I cannot recommend a better person to follow than Matt Levine. He's got a bachelor's in Classics from Harvard (which shows when he breaks out grammatically-correct Latin jokes), once opened a post about Federal Reserve interest rates with a parody of the Saint Crispin's Day speech from Henry V, and refers to biotech startup Theranos as "Theranos, the Blood Unicorn" because, well, it's a blood-testing startup worth $1B it's funny.

Some of it is just gold:

Behold:

The board (the “Board”) of directors (the “Directors”) of the Company wishes to inform the shareholders of the Company (the “Shareholders”) that on 4 December 2015, a truck of the Group (the “Truck”) loaded with, among other things, all original financial documents of the Group for the four financial years ended 31 December 2014 and for the current year (the “Lost Documents”) were stolen in the Qingyuan District of Baoding City, Hubei Province, China while the truck driver was taking a lunch break on his journey to transport the Lost Documents back to the Group's head office in Beijing (the “Incident”).

If you read a better sentence than that this week, you are very lucky. The defined terms alone -- "(the 'Truck')," "(the 'Lost Documents')," "(the 'Incident')" -- are pure poetry, but the narrative flow of the sentence -- starting with a banal corporate setup, foreshadowing with the ominous "among other things," slipping the ungrammatical climax ("were stolen") in between irrelevancies (the fiscal years of the documents, the precise neighborhood of the theft), and finally lightening the mood with the quotidian comedy of the driver's lunch break and the crowning absurdity of "the Incident" -- is so compelling that I recommend that you not read the rest of the press release. (...)

A lot of it, unfortunately, is only maximally funny if you've worked in financial services, like this utter geekery about the Fed rate hike:

In a sense -- in a real, economically meaningful sense -- the Fed has already raised rates, because we have talked about it so much. In a sense, the Fed decision takes place within us, each day. In another, more accurate sense, though, "the Federal Open Market Committee releases its policy statement at 2 p.m. Wednesday," and will likely announce that it is raising short-term interest rates above zero for the first time in seven years. (...)

Or this story about Yahoo!, Inc., which is in some sense valued at negative thirteen billion dollars because the markets don't trust it to not mess up the plan to sell some stock it owns properly. (Granted, "properly" here actually means "sell everything that isn't the stock to Yahoo 2.0, then sell Yahoo 1.0", because, well, the "some stock" in question is worth more than Yahoo itself.)

If you want to spend some time hitting your head against the table in incredulity, try this story about twin trader scammers whose "alleged scam consisted more or less of the twins pretending to be each other." Writes Levine: "Come on! That is the plot of a teen comedy, not a securities enforcement action."

And finally, if you want to learn something, here's a well-written (if technical in parts) piece on how just making the system slower won't make high-frequency trading go away.

Also, hat tip to Matt: The NYT prints Hillary Clinton and Bernie Sanders's plans to "rein in Wall Street", and economist Larry Summers responds to Sanders's in the Washington Post. I have to say, Summers makes a much, much better economist than he does university president, and I'd even call his response "definitely worth reading".

Well, that's half the blogs (and more than half of the links) I've got from December. Stay tuned for Part II, including medicine, lifehacks, effective altruism, and anything that didn't come from a regular blog.