notable thought vectors

  • thought vector (ML n.): digital representation of an idea; a series of numbers produced by embedding e.g. the words "cow", "heifer", and "Aberdeen Angus" in several languages, but also (speculative) all these images.

    if you can convert each sentence in a document into a vector, then you can take that sequence of vectors and [try to model] natural reasoning. And that was something that old fashioned AI could never do.

    If we can read every English document on the web, and turn each sentence into a thought vector, you've got plenty of data for training a system that can reason like people do. (Now, you might not want it to reason like people do, but at least we can see what they would think.)

    What I think is going to happen over the next few years is this ability to turn sentences into thought vectors is going to rapidly change the level at which we can understand documents.
    A big ass deal.

    Of course, it is not computationally possible to deal with than a couple of thought vectors at once. Currently. And we'd need many thousands to match our brightest exegetes. But if it could be done, it would be a implementation / proof of the Tractarian philosophy.

    (Geoff Hinton has a few of these giant hints about the future (see also dark knowledge, a residue function extracted from a massive ensemble model; can perform efficiently, even on data not mentioned in its class probabilities, even on smartphones).)

  • clone (trans* pej. n.): Insult for a trans person who seeks to 'pass' perfectly. Seems pretty nasty, but I've seen a non-pejorative use. See also fish, Mary Kay, GG.

  • to hellban (internet v.): to make posts by a user invisible to all other users, without telling them; to covertly enforce ostracism. A taste of death. Psychologically sound: "When nothing they post ever gets a response, a hellbanned user is likely to get bored and leave" rather than just creating a new account. Very malign despite its effectiveness.

  • clapback (US n.): a comeback; rapid response to insult. From the old-school rap thing of syncopating your sentence with claps? And 👏 so 👏 it 👏 explains 👏 this 👏 annoying 👏 meme.

  • wypipo (US pej. n.): Racial slur; white people. (But it doesn't count as a slur: demeaning and upsetting people isn't real, systemic harm, after all.)

  • call girl principle (n.): Rule for negotiating: set the price / collect payoff before rendering anything. This is because the perceived value of a service is greatly diminished afterward. Anchoring, hindsight bias, post-purchase rationalisation: it's all in there.

  • Superfund site (US n.): a highly toxic hellhole, suitable for receiving emergency government cleanup funds.

  • to stovepipe (software, pej. v.): to build separately, without integrating into other systems. A "stovepipe system" could share data with other systems but doesn't. (Consider one chimney per house in a terrace of houses.)

  • toaster (software, n.): 1) of an inferior computer, one overheating as it tries to keep up with modern software demands. 2) freeware consisting entirely of glued-together libraries.

  • lagniappe (Fr. n.) from yapay (Quechua): a little extra. Used particularly in New Orleans. See also Irish English luckpenny.

  • calaboose (n.) from calabozo (Spanish n.): prison, particularly a small one.

  • modifiable areal unit (geography n.): Cute term for something which doesn't exist in nature, which is only conventional or quasi-real. Some parts of philosophy are unavoidable after all.

  • Whispernet (n.): The free global wi-fi that Kindles get. Kinda limited, but still cool and ekeing a little more wonder out of this 'knowledge pulled out of the sky' thing we are given without effort.



My nation is a dress uniform, like all nations. Distinctive, colourful, old, mass-produced. Six sizes too big. If I wear it I am legible to you, you I haven't met. It veils me when I meet you and don't want to be met by you. It lives in the cupboard (I don't have to meet many people).

It's usually nice to own it - something to don when surrounded by notional barbarians, to set myself apart in my different barbarism. Though often people point to it, saying that I am my frock coat, or that I'm wearing my frock coat when I'm not wearing it. This isn't fun, as no forced game is fun. At least my coat isn't caked in shit and blood, like yours. (Like all coats, it is caked in shit and blood, but at least mine isn't on the outside.)

Like all regiments, my regiment thinks it is special: not many people have these coats. But wearing any coat makes you less rare: you leave your kingdom singly for a low foothold on Leviathan.

No one will spit on my coat, unlike yours. It is humble and demotic. The regiment's crimes were quiet crimes, or else loud crimes with none left to say.

Frock coats are new. There are no frock coats, we just pretend we're wearing frock coats to humour each other. It is cold without them. But it it wouldn't be, if you stood like so before yours as it burned. Say, can you see?


thole thule

Pale rulered ceiling low.
The fog a second sky at ten paces.
Your breath a third foglet.
You'd review harshly a film ending
among this melodramatic a cloud chamber,
lazy with meteorological ellipsis.

Away, you forget endmost
Grampian, the uniformity and wall-eyed mist.
Back, grey cries for colour: quayside tattoos,
neon dye, Jäger. Colour isn't given.

Nae thermo, nae sae dynamic. (Ootsides, onywauy.)
Folk thole the grey reef lang enou,
puddle in the sea, hoovering
at livid macroeconomic cracks.
Abdy oxidates, no white-het but blue.
A'hin blurs. A'hin levels. A'hin mixes. A'hin cools.


notable signage, postage, and oral topiary

  • incel (n.): involuntary celibate. Tends to be used by terribly bitter people. Ozy Frantz is trying to reclaim the term, since the problem is a real one.

  • foofaraw (West US n.): pomp, fuss, ruckus.

  • sinter (metallurgy v.): to fuse things into a solid without heating them to liquefaction. Think making a vicious snowball out of powder. Useful for ultra-high-melting-point things like tungsten.

  • zero-rating (telecomms n.): Offering free mobile data, usually for very limited things like app stores or Facebook. A good form of price discrimination - but legislators often ban it in the name of net neutrality, creating equality by running among fields with a running chainsaw. Common in the developing world - best of all, for Wikipedia.

  • dead mileage (n.): non-profitable movement of commercial vehicles, for maintenance or live placement.

  • to deadhead (v.): to travel, as staff, using dead mileage.

  • data sleaze (n.): customer data obtained secretly by businesses, secretly sold on. Almost all 'free' services are data sleaze operations.

  • expert beginner (pej. tech n.): Closed-minded person, who plateaus before becoming actually competent, because of bad feedback or arrogance. Often hailed as an expert by other know-nothings in their small pond. Most very experienced people are probably this. ("10 years of experience... or the same year repeated 10x?")

  • ignotum per ignotius (Lat. n.): An explanation which is more obscure than the initial concept. e.g. In which an explicand, which is to say one or more explananda, is elucidated by a kaleidoscopic but wholly inefficacious explanans, the whole taking on a manner befitting only an ἐσωτερικ or γνωστικ, thereby violating the good maxims of relation and manner.

  • language server: 1) the part of an IDE that supplies autocomplete and refactor-all and all that sugary goodness; 2) a good name for a blog.



How many hours have I spent on maths?

(or, rather, How many hours have I spent explicitly doing maths?)
  1. Formal education
    • Preschool. 300 hours of counting?

    • Primary school. (4? per week * 40 weeks x 7 years) = 1100

    • Secondary school (6 years)
      S1+S2: (3? per week * 40 weeks * 2) + homework (80), revision (40) = 360
      Standard Grade: (3? per week * 40 weeks * 2) + homework (160), revision (50) = 450
      Higher Maths: (3 per week * 40) + homework (100), revision (100) = 320
      Physics: (2 * 40 * 1) + home (50) + revision (50) = 180
      Chemistry: (2 * 40 * 1) + home (50) + revision (50) = 180
      =~ 1800

    • Higher education
      Economics = (4 hours * 24 weeks * 4 years) + homework (200) + revision (200) + thesis (100) = 900
      Cryptography = 100
      Half a BSc in Maths = 16 hours per assignment x 23 assignments + (60 hours study x 4 courses) = 600
      ~ 1600

    ~ 5000 hours.

  2. Researches
    My own writing hasn't been very mathematical so far. But I've done a bunch of recreational bits and warmup pieces, and read books like GEB and Chaos.
    ~ 200 hours.

So I am supposedly halfway to Gladwellian mastery. Sure doesn't feel that way; but then, the first two-thirds of those hours were undermined by their being coercive and applied to a formless fog of feeling, me as a child. And I have scarcely ever been 'deliberate' about it.


notable nah

  • threeper (US n.): A 'Three Percenter', member of a post-Obama movement of private militias. Mostly hunting Mexicans along the border and fuming about what Alex Jones tells them about the government. 'Obama as globalist', 'blue hats invading', 'sheeple', 'chemtrails': it's all in there.

  • screecham (Shetland n.): whisky. See also skreigh, from 'shriek'.

  • Terminus (Lt. proper n.): Roman god of boundaries. This made me laugh: His statue was merely a stone or post stuck in the ground to distinguish between properties..

  • kindness sandwich (Lt. proper n.): Gentle criticism, tempered by putting the corrective inbetween two compliments. See also bumblebee nickname ("Cute, but it stings.")

  • AFOL (n.): Adult Fan of Lego.

  • veg (gamer pej. adj.): vegetative; of wasting time on games. To veg out is established British slang, but the adjectival form is new to me: "You've got a really veg profile."

  • egofag (4chan pej. n.): An attention-seeking person. See also dramallama.

  • algorism (n.): the practice of arithmetic using (Indian) positional numbers. Took 400 years to really catch on, after initial import to Europe.

  • new god argument (n.): The claim that belief in God ('superhumanity') is justified on simulationist and Hegelian transhumanist grounds. 'You should believe in God because we will become God.' But more interesting than that implies. Original work by a Mormon dude with an incredible name.

  • to krige (v.): to interpolate values using a Gaussian process regression. AKA Wiener-Kolmogorov. "kraɪg" I think.

  • Chi-Town (US n.): Chicago. Pronounced "Chai".

  • bitext (n.): a parallel text (writing in two languages, side by side).

  • rebar (n.): reinforcing bar: the cool ropey steel in reinforced concrete.

  • Paremiography (n.): the study and collation of aphorisms.

  • inducer (n.): A worse name for 'learner', learning algorithm. From (logical) induction.

  • philosophy (n.): arguing about definitions, arguing about implications, arguing about possibilities, without being able to check your working, or assign probabilities.

  • evolution (n.): an agent-based iterative search over phenotype space.

  • machine learning (n.): or pattern recognition, statistical learning theory, data mining, knowledge discovery, predictive analytics, data science, adaptive systems, self-organizing systems.


Fudge factors for military PR

  • The Geneva ratio: how many deaths by conventional weaponry does it take to equal the outrage of a single death from chemical weapons? (Around 10,000)

  • The Matthew 18:6 ratio: how many adults can die tragically before they register as much as one child's death does? (More than 1000.)

  • The robot ratio: how many deaths can a manned aircraft cause before it equals the outrage accruing to a single drone strike? (Around 10?)

  • The terrifying terrestrial terror ratio: how many people can be killed in e.g. car accidents before the outrage of a single killing by a terrorist is matched? (More than 1000.)

  • The Vertreibung ratio: how many civilians of an enemy state can be killed before the outrage of a home civilian's death is matched? (1 million divided by 2400*...)

  • The fatigues: How many home-nation soldiers can die before the outrage of a single home-nation civilian death is matched? (In the dozens?)

Solve for max casualties and min retweets.

(This post was originally titled "Fudge factors for Public Affairs" but this closely related term is too obscure.)

(I also resisted the urge to make this stupid pun:
  • The Krypton factor: How many people can Lex Luthor kill before he matches the outrage of Superman letting a single person die?

* 2400 military casualties, true, but they weren't at war and so borrowed some of the civilian halo for a moment.


if by notable you mean you mean the devil's brew, the poison scourge, the bloody monster, that defiles innocence, dethrones reason, destroys the home, creates misery and poverty, yea, literally takes the bread from the mouths of little children; if you mean the evil drink that topples the Christian man and woman from the pinnacle of righteous, gracious living into the bottomless pit of degradation, and despair, and shame and helplessness, and hopelessness, then certainly I am against it.

  • fakery (US n.): a bakery that orders frozen pastries from a distributor while presenting them as fresh. Surely most bakeries are by now.

  • en banc (Fr. / legal adj.): "In bench": heard before every judge in the court at once. Marker of severity.

  • to toggle into memory (1950s v.): to enter a program into an old mainframe, one bit at a time, using the switches on its front panel.

  • to vernalize (v.): to induce a plant to flowering by exposing it to prolonged cold (a la winter).

  • stim toys (n.): stress balls and widgets and fidgets that allow healthy displacement behaviour for autistic people in social situations.

  • outsider porn: literary novels.

  • pandonic (programming adj.): of idiomatic Pandas code.

  • to perp (v.): to perpendicularise. Term in geometry / computer graphics for effecting a 90 degree rotation with simple swap- and negate logic.

  • vidya (US pej. n.): 1) computer game. I got excited for a moment when someone said that teenagers were all really into 2) vidya.

  • randomista (n.): One who champions the use of RCTs in the formulation of policy, especially poverty policy.

  • Magic Circle firm (n.): One of four very prestigious London law firms, all among the highest-revenue in the world. "Set for life."

  • backpressure (n.): Crashing behaviour of gases in kinked pipelines. 1) 2) a Microsoft Mailbox server feature that queues up messages when load is too high.

  • memsec (n.): 'memory-second': the unit of consumption of cluster resources: 1MB RAM filled for 1 second.

  • Dead Sea effect (n.): the phenomenon by which a minor policy mistake in a IT shop causes long-term degradation of the staff: good people are more mobile, so they leave, so the work , so more good people leave, so meh people are left, so the shop bears a cyclical burden of mediocrity: hard to get good people in, and no one great stays. Coined by Bruce Webster.

  • the Surinder Singh route (n.): Circumventing British anti-immigration law by having your non-EU spouse live in the EEA for a bit, which opened up an alternative legislative route to UK citizenship by marriage. Named for the plaintiff who established the precedent. Will go soon, of course, but I love little legal easter eggs like these.

  • TEPES (n.): your talent, education, professionalism, experience, skills. Another good Bruce Websterism.


Been reading, Q1 2017

(c) Woman Reading (1970) by Wil Barnet

(c) The Times, 1849

I have never focussed on any thing for very long. In 2008 I took advantage of my government's inefficient good graces to study philosophy, thinking myself serious. In 2010 I started a blog, thinking myself serious. In 2012 I started reviewing everything I read, thinking myself serious. In 2014 I learned to code, and thought myself very serious. In 2016 I got myself a good, hard job and continue to think myself serious. But my job involves induction.


  • Sex by Numbers (2015) by David Spiegelhalter. Very fun, but with a serious scientific mission. The expected titillating facts are of course present (how many people have tried anal? How many people are gay? What's typical?) but there's also an intro to the many difficulties of social science, and a history of sexology in here. You learn why you should admire but not trust Ellis, Hirschfeld, Kinsey, Masters & Johnson, Hite... Something for everyone.


  • Evolving Ourselves (2015) by Steve Gullans and Juan Enriquez. Broad-minded venture capitalists seek to update Darwinism in light of new human capabilities. 100 tiny chapters on some facet of modern genetics and modern genomes and epigenomes and microbiomes... Topics are incredibly varied and excited, but it's sugary.


  • Chaos (1989) by James Gleick. Romantic, dramatic, and genuinely additive pop science: the physics, meteorology and maths in this was famed but not well-explained before this came out. The theme of the very different results presented here is unprecedented successes in recognising and explaining nonlinear systems. Very human: every researcher is profiled sensitively, generally as an outsider challenging the stuffy, desk-bound precepts of 'linear science'. Since ornery, heroic Mandelbrot is included here, you get an exciting ride even if you don't like maths or science or the world or the underlying generative process of all instances of beauty.

    "Chaos" is a bad name for the field: it implies randomness, indeterminism, intractability. Better would be to question why the word "order" can only refer to 1) equilibrium or 2) periodic patterns - why it is we think of order as boring. "Deterministic disorder" is more honest - and better yet is Lao-Bin's "order without periodicity".

    Also, the diagrams are poor by contemporary standards: I had to stare at them for a while before grokking the concept of them.

    Borne on what must have felt like an epochal wave, Gleick overreaches. He calls Smale and Mandelbrot "the end of the reductionist program in science". How is seeking and finding a precise (nonlinear) equation - which is the case in the work of all these men - for a system holist!? I don't actually know if the maths in here has changed everything: maybe it has, and they suffer from the Seinfeld effect for dynamical systems, seeming obvious after the fact. But I do know that the Santa Fe strain of work is more of a tolerated eccentric uncle than a science-upending behemoth.


  • Age of Em (2016) by Robin Hanson. A truly remarkable book; easily in the top 5 most insightful out of the 400 I have reviewed here. Last year I called Superintelligence the most rigorous exploration of the nonreal I had ever read: this beats it by a lot. You will find yourself reading pages on the properties of coolant pipes and be utterly engrossed.

    People tend to wrap this book in ulterior motives and esoteric intentions, because they love it but see futurism as an unworthy goal for such an achievement. I am no different: this is the greatest compendium of real social science I have ever found.

    No review can do much justice, but here's one particularly hair-raising point in it: Hanson surveys the whole course of human history, and notes the many ways our culture is unprecedented and, in the evolutionary sense, nonadaptive:
    we live in the brief but important “dreamtime” when delusions [drive] history. Our descendants will remember our era as the one where the human capacity to sincerely believe crazy non-adaptive things, and act on those beliefs, was dialed to the max.

    Why is our era so delusory?
    1. Our knowledge has been growing so fast, and bringing such radical changes, that many of us see anything as possible, so that nothing can really be labeled delusion.
    2. Rich folks like us have larger buffers of wealth to cushion our mistakes; we can live happily and long even while acting on crazy beliefs.
    3. We humans evolved to signal various features of ourselves to one another via delusions; we usually think that the various things we do to signal are done for other reasons. For example, we think we pay for docs to help our loved ones get well, rather than to show that we care. We think we do politics because we want to help our nation, rather than to signal our character and loyalty. We are overconfident in our abilities in order to convince others to have confidence in us, and so on. But while our ancestors’ delusions were well adapted to their situations, and so didn’t hurt them much, the same delusions are not nearly as adapted to our rapidly changing world; our signaling induced delusions hurt us more.
    4. Humans seem to have evolved to emphasize signaling more in good times than in bad. Since very few physical investments last very long, the main investments one can make in good times that last until bad times are allies and reputation. So we are built to, in good times, spend more time and energy on leisure, medicine, charity, morals, patriotism, and so on. Relative to our ancestors’ world, our whole era is one big very good time.
    5. Our minds were built with a near mode designed more for practical concrete reasoning about things up close, and a far mode designed more for presenting a good image to others via our abstract reasoning about things far away. But our minds must now deal with a much larger world where many relevant things are much further away, and abstract reasoning is more useful. So we rely more than did our ancestors on that abstract far mode capability. But since that far mode was tuned more for presenting a good image, it is much more tolerant of good-looking delusions.
    6. Tech now enables more exposure to mood-altering drugs and arts, and specialists make them into especially potent “super-stimuli.”... today drugs are cheap, we can hear music all the time, most surfaces are covered by art, and we spend much of our day with stories from TV, video games, etc. And all that art is made by organized groups of specialists far better than the typical ancestral artist.
    7. We were built to be influenced by the rhetoric, eloquence, difficulty, drama, and repetition of arguments, not just their logic. Perhaps this once helped us to ally us with high status folks. And we were built to show our ideals via the stories we like, and also to like well-crafted stories. But today we are exposed to arguments and stories by folks far more expert than found in ancestral tribes. Since we are built to be quite awed and persuaded by such displays, our beliefs and ideals are highly influenced by our writers and story-tellers. And these folks in turn tell us what we want to hear, or what their patrons want us to hear, neither of which need have much to do with reality.

    These factors combine to make our era the most consistently and consequentially deluded and unadaptive of any era ever. When they remember us, our distant descendants will be shake their heads at the demographic transition, where we each took far less than full advantage of the reproductive opportunities our wealth offered. They will note how we instead spent our wealth to buy products we saw in ads that talked mostly about the sort of folks who buy them. They will lament our obsession with super-stimuli that highjacked our evolved heuristics to give us taste without nutrition. They will note we spent vast sums on things that didn’t actually help on the margin, such as on medicine that didn’t make us healthier, or education that didn’t make us more productive.

    Our descendants will also remember our adolescent and extreme mating patterns, our extreme gender personalities, and our unprecedentedly fierce warriors. They will be amazed at the strange religious, political, and social beliefs we acted on, and how we preferred a political system, democracy, designed to emphasize the hardly-considered fleeting delusory thoughts of the median voter rather than the considered opinions of our best experts.

    Perhaps most important, our descendants may remember how history hung by a precarious thread on a few crucial coordination choices that our highly integrated rapidly changing world did or might have allowed us to achieve, and the strange delusions that influenced such choices. These choices might have been about global warming, rampaging robots, nuclear weapons, bioterror, etc. Our delusions may have led us to do something quite wonderful, or quite horrible, that permanently changed the options available to our descendants. This would be the most lasting legacy of this, our explosively growing dream time, when what was once adaptive behavior with mostly harmless delusions become strange and dreamy unadaptive behavior, before adaptation again reasserted a clear-headed relation between behavior and reality.

    Our dreamtime will be a time of legend, a favorite setting for grand fiction, when low-delusion heroes and the strange rich clowns around them could most plausibly have changed the course of history. Perhaps most dramatic will be tragedies about dreamtime advocates who could foresee and were horrified by the coming slow stable adaptive eons, and tried passionately, but unsuccessfully, to prevent them.
    It's easy to read a radical critique of our liberal values in there, but I believe him when he says that he doesn't dislike dreamtime; he just predicts it cannot last, because we are fighting an old and inexorable tide.

    There are several thoughts this large, and a thousand other small insights in Age of Em.


  • The King James Bible, Genesis (1611) by the First Westminster Company. Doing a chapter a day. (I only realised afterward that this is 3 years' labour.) Prose really is uniformly good, fresh - even bearing the weight of all the bizarre convolutions in the mind of the original authors ("in his kind, and in his kind and in his kind").


  • 80,000 Hours
    (2016) by Benjamin Todd et al. Collation of results from a very grand project: to channel young careerist thousands into better tasks in higher gear. If you have the will to do well you should read the website, and think through the planning exercise here. Unlike everything else I've read about career development, since it talks about work and success without being nauseating bullshit.

    4*/5 for anyone under 40.
    [Same material free here]


  • Humanity's Burden: A Global History of Malaria (2008) by Webb. Worthy, thoughtful, and on one of the most important topics in the history of the world. Didn't know that malaria was one of the many curses of the Columbian Exchange: it wasn't even on the continent before us. It was, however, prevalent in the marshes of Essex. Not useful per se, but it gives you a sense of the size and ancestry of the beast we are hunting.


  • (One of the most hideous book covers I've seen btw.)

  • The Quest For Artificial Intelligence (2010) by Nils Nilsson. A sweet informal history of AI research from a Stanford doyen. In places it is oral history -
    ...Jack was the Director of DARPA from 1987 to 1989 and presided over some cutbacks in AI research (including the cancellation of one of my own research projects)
    Like any history, the history of computing is full of little myths - e.g. that Lovelace was the first programmer, that von Neumann originated stored-program memory, that ENIAC was the first true computer, that hardware and software is a clean and natural division in kind... Nilsson calmly lets out the air of these and more.

    [Free here]

  • Reread: The Gigantic Beard that was Evil (2010) by Stephen Collins. Gorgeous, but not as deep as I felt last time.



  • Out of Sheer Rage: in the Shadow of DH Lawrence (1997) by Geoff Dyer. A book about an unwritten book about a writer I don't like much. And it's amazing! Not a study of Lawrence, a study of trying to write when you lack an iron will. So also a study of all work, so a study of the hard generation of value, and so, despite appearances, a study of what matters.

    The prose is circuitous, cantatory, shaggy-dog, but never dull:
    Oxford! Now if there is one place on earth where you cannot, where it is physically impossible to write a book about Lawrence it is here, in Oxford. You could write a book about plenty of writers in Oxford: Hardy, or Joyce even — people are probably doing just that, even now, dozens of them — but not Lawrence. If there is one person you cannot write a book about here, in Oxford, it is Lawrence. So I have made doubly sure that there is no chance of my finishing my study of Lawrence: he is the one person you cannot write about here, in Oxford; and Oxford is the one place where you cannot write about Lawrence.

    When I say you can’t possibly write a book about Lawrence in Oxford that is not to be taken too literally. At this moment, within a few miles of my flat, dozens of people are probably writing books about Lawrence. That tapping I can hear through my open window is probably someone writing a book or a thesis or preparing a lecture, or, at the very least, doing an essay on D. H. Lawrence. It can be done. It can be done — but it can’t be done, it shouldn’t be done. You can’t write a half-decent book about Lawrence in Oxford, can’t write any kind of book about Lawrence without betraying him totally. By doing so you immediately disqualify yourself, render yourself ineligible. It is like spitting on his grave.
    For a while I amused myself by seeing how many consecutive sentences used the same phrase, in a running stitch motif. He is playing a character, but like Rob Brydon does: only slightly heightened.

    One long stream of scenes, unthemed, unbracketed. He is the critic I would have hoped to be: sceptical of the novel, sceptical of the spiritual pretensions of artists, sceptical of children, sceptical of travel and sceptical of home, sceptical of self. He is free to admit his boredom and his joy, unlike the academic critics he often erupts against. Here is the key passage (not that you can trust him to cleave to it twenty years or minutes on):
    Hearing that I was ‘working on Lawrence’, an acquaintance lent me a book he thought I might find interesting: A Longman Critical Reader on Lawrence, edited by Peter Widdowson. I glanced at the contents page: old Eagleton was there, of course, together with some other state-of-the-fart theorists: Lydia Blanchard on ‘Lawrence, Foucault and the Language of Sexuality’ (in the section on ‘Gender, Sexuality, Feminism’), Daniel J. Schneider on ‘Alternatives to Logocentrism in D. H. Lawrence’ (in the section featuring ‘Post-Structuralist Turns’). I could feel myself getting angry and then I flicked through the introductory essay on ‘Radical Indeterminacy: a post-modern Lawrence’ and became angrier still. How could it have happened? How could these people with no feeling for literature have ended up teaching it, writing about it? I should have stopped there, should have avoided looking at any more, but I didn’t because telling myself to stop always has the effect of urging me on. Instead, I kept looking at this group of wankers huddled in a circle, backs turned to the world so that no one would see them pulling each other off. Oh, it was too much, it was too stupid. I threw the book across the room and then I tried to tear it up but it was too resilient. By now I was blazing mad...

    I burned it in self-defence. It was the book or me - writing like that kills everything it touches. That is the hallmark of academic criticism: it kills everything it touches. Walk around a university campus and there is an almost palpable smell of death about the place because hundreds of academics are busy killing everything they touch. I recently met an academic who said that he taught German literature. I was aghast: to think, this man who had been in universities all his life was teaching Rilke.
    Rilke! Oh, it was too much to bear. You don’t teach Rilke, I wanted to say, you kill Rilke! You turn him to dust and then you go off to conferences where dozens of other academic-morticians gather with the express intention of killing Rilke and turning him to dust. Then, as part of the cover-up, the conference papers are published, the dust is embalmed and before you know it literature is a vast graveyard of dust, a dustyard of graves. I was beside myself with indignation. I wanted to maim and harm this polite, well-meaning academic who, for all I knew, was a brilliant teacher who had turned on generations of students to the Duino Elegies. Still, I thought to myself the following morning when I had calmed down, the general point stands: how can you know anything about literature if all you’ve done is read books?

    Now, criticism is an integral part of the literary tradition and academics can sometimes write excellent works of criticism but these are exceptions - the vast majority, the overwhelming majority of books by academics, especially books like that
    Longman Reader are a crime against literature.
    The final passage hits you over the head with what you have certainly already worked out, but it is still very powerful. Dyer is inspiring, pure nevertheless:
    One way or another we all have to write our studies of D. H. Lawrence. Even if they will never be published, even if we will never complete them, even if all we are left with after years and years of effort is an unfinished, unfinishable record of how we failed to live up to our own earlier ambitions, still we all have to try to make some progress with our books about D. H. Lawrence. The world over, from Taos to Taormina, from the places we have visited to countries we will never set foot in, the best we can do is to try to make some progress with our studies of D. H. Lawrence.
    4*/5. (Once only.)

  • Stories of Your Life (2002) by Ted Chiang. Astoundingly good; he is our Borges. The stories are miscellaneous, but all bear the weight of one core theme - that rationalism, materialism is not the enemy of humanism, but is much more able to accommodate us, our highest values, than is romantic supernaturalism.

    So he's an artistically successful Yudkowsky; Chiang's own presumable nerdiness disappears in his powerful but austere prose, even when characters are expounding the principle of least action or the details of ancient masonry.

    'Story of Your Life' is so much more interesting, emotionally and scientifically, than the Arrival film it was made into. 'Tower of Babylon' is rousing minutiae. 'Hell is the Absence of God' takes the tired, speculative, stupid themes of the Abrahamic conversation - faith, will, love, persistence, atheism - and wrings out a new chord from them. Ah!



Done in March 2017

  • Archived webpages cited on my sites. Pride: 2.
  • Posh Spark training. Pride: 1.
  • Maths assignment #6: vector calculus, contours, fields. Pride: 3.
  • Started functional programming (Odersky, Chiusano). Pride: 2.
  • 15 posts not including this. Pride: 3.
  • Read Chiang, Dyer, Webb, Stross. Pride: 1.
  • Took the lead hiring. Pride: 3.
  • Lots of plans for software and datasets. Pride: 0?


notable emanations

  • WORM (adj.): Write Once, Read Many (times). A particular use pattern in databases, that allows for faster but less safe NoSQL systems.

  • bitterant (n.): bitter chemical, applied to things like antifreeze to stop children and other animals eating them.

  • bruxism (n.): excessive grinding of teeth. Apparently it is not painful, but still horrendous to me.

  • relevance engineer (n.): A data scientist focussing on search or recommendation. (I note keenly the shifts in terminology which imply any devaluation of the hot and silly coin in my pocket, "data science"; on the recent StackOverflow survey a large chunk of people responded "machine learning specialist".)

  • incentive compatibility: One of the most important ideas in economics, but I never learned the name in 4 years of indifferent study. "I don't agree with him at all, but we are incentive-compatible."

  • to die in harness (v.): Melodramatic version of death in service. It is the official term in Indian government, inherited from the Raj.

  • data room (n.): an exhaustive dump of backroom information about a company. Assets, cap table, contracts, disclosures.

  • metis (n.): nonpropositional knowledge; hunches, feels, heuristics, Simonian bon sens. If intelligence is the trait that leads to knowledge (scientia), phronesis leads to metis. Rationality leads to wins.

  • ganzfeld study (n.): An experiment that tries to detect ESP in a particular science-esque way.

  • gentleman ranker (n.): a disgraced, discharged officer who re-enlists as a private. But see also the case of John Hume Ross.

  • remittance man (n.): a disgraced scion, sent off to Canada or Australia or Morocco, and sent a monthly stipend just so long as he doesn't come back.

  • memory pressure (n.): in distributed systems, the risk introduced by maxing out allocation.

  • data enrichment (n.): just adding more columns. But expensively.


incentive compatibility

We live together: you dislike mess more than you dislike cleaning; I dislike cleaning more than I dislike mess. Mess happens: obligate social grooming rears a silent scowling face.

A current account runs to deficit: cogwheels backlash. I could offer you money for doing my part, if I was stupid; or if you were a different species. As you are this is a grave insult: cleaning you undertake yourself is home-making, comfort behaviour, preening, an act conceived in freedom and ease. Receiving money for it makes you a cleaner: low-status. Offering you money called you low status: I signalled superior wealth. Negotiations sour: you don't hear my offer instead to cook, or do the bins. But we are grown men; there must be a solution.

Yes: I skip the lease under cover of night, free-riding the axle of a Scania bound for the orient. In the morning: notes stuffed under your door.


standard repertoire in computer science

Don Knuth at his home organ

A trained musician knows hundreds of pieces, many of them from "standard repertoire", a list of classics. I'm getting by in a statistical / computer scientific career, despite not being trained per se in either. What's in the computer science / IT / hacker repertory?

Here is a list of things you should be able to use or define by the end of a good computer science undergraduate degree. (Starred are ones which will enhance your life most, whether with hundreds of thousands of pounds, or hundreds of hours, or a larger practical-ethical expansion. Career improving ones are pretty obvious, but in computing the divide between the fascinating or improving, and the employable, is narrower than elsewhere.)

  • Get into DIY, in its grand philosophical sense: create, not just consume.** Portfolio, not resume!
  • Demystification of tech.**
  • ...and thus participation in the defining activity and mindset of the age.*
  • ...and thus scepticism about the many expensive and ludicrous parts of it.**
  • The imaginative leap of non-WYSIWYG working. Real mental modelling.
  • Appreciation of the uniqueness of programming as tool in any inquiry.

  • Hardware assembly and maintenance.*
  • Overcoming your fear of the command line (*nix preferably).
  • Thus scripting to automate dumb (and smart) stuff.*
  • Install, configure, compile Linux.
  • Compile and configure a web server.
  • Compile and configure a DNS daemon.
  • network protocols and socket level programming.
  • monitoring, reporting, fail overs etc.


A fraught topic, for some reason. No particular language is indispensable, but there are at least 10 important axes to understand.
  • For development speed: Python or Ruby
  • For execution speed: C or Rust or Go
  • For portability: Java or C or bash
  • For nostalgia or ritual: C
  • For elegance: Haskell
  • For puzzles: Prolog or uKanren

  • An Algol ("C-like"): Java or C#
  • A Lisp: Racket or Scheme
  • An ML: F# or Haskell

  • For data abstraction (Assembly or anything)
  • For class abstraction (C# or Java)
  • For type abstraction (Python or Javascript)
  • For functional abstraction (Haskell)
  • For syntactic abstraction (Scheme)
  • For implementation abstraction (SQL or Prolog)
  • For processor abstraction (Erlang or Go)

  • Standard web markup: HTML/CSS/XML/JSON
  • Markdown for rapid writing.
  • Regex for fast text manipulation.*
  • LaTeX for beautiful technical writing.
  • A version control system.

  • Theory of computation. (Why are Turing, Shannon, von Neumann among the greatest thinkers of the past hundred years?)
  • Theory of computational complexity.

  • hash table
  • linked lists
  • trees
  • directed and undirected graphs
  • binary search trees

  • compiler, linker, interpreter: use and theory of
  • virtual memory and paging.
  • kernel mode vs. user mode
  • threading
  • synchronization primitives
  • platform internals: disassemblers, decompilers, debuggers...
  • The entire programming stack:
    • hardware (CPU + Memory + Cache + Interrupts + microcode)
    • binary code, assembly
    • static and dynamic linking
    • JIT compilation
    • garbage collection, heap, stack, memory addressing

  • number theory for crypto
  • automata
  • formal grammars

  • The Nature of Computation
  • Code Complete 2
  • Don't Make me Think
  • Design Patterns
  • Peopleware,
  • Programming Pearls,
  • The Pragmatic Programmer,
  • Mythical Man month
  • Structure and Interpretation of Computer Programs,
  • Art of Computer Programming
  • Hacker's Delight

to be led out

You start to learn something. You don't know what to google. You don't know the luminaries. You don't know what are stupid questions. You don't know which are the good books, and they are all £90. You might not know a good book when you saw it, except that it seems to make sense to you, where others are demeaning slammed doors. You don't know enough to just get started and incrementally improve at any rate at all.

Education is artificial enclosures and screens on this terrifying commons: ignore those cliffs, forget that vertiginous sky, stay in here, you will be safe to get strong, here is a nice story. StackExchange is a chain of lifeboats on the open sea of research, vanishing to the horizon.

Most educated people never leave the enclosures, and mistake the limits of the curriculum for the limits of the world. (In this way, it's possible that the American general education philosophy - so admirable, so civic - could narrow minds.) In economics this "101ism" is particularly pernicious, since even honest specialists, those operating well beyond the screen, can't communicate their technical results to the media, so almost all discourse takes place inside the fake, narrow enclosure, with endless fruitless illiberal results.

In fields where it's impossible to know if you have gone astray - everything except the formal sciences - the work feels nicer but is sadder, considered on a proper timescale, of centuries. There it is almost inevitable that lives will be ploughed into the soil and merge with the stream of decomposing misguided theses. In the formal sciences this is only very likely.


notable labiodentals

  • E/N site (n.): An "Everything/Nothing" site. (As in, "means everything to the person who's writing it, and nothing to everyone else".) The 90s word for blog.

  • Greek life (US n.): Amusing Ivy League slang for the frat / sorority system. Gives rise to excellent shite like this:
    ...After nearly three decades of operating in the shadows, the Greek organizations could find themselves under the University's regulatory oversight or banished altogether, President Tilghman said May 5.

    "At the moment I am keeping an open mind about all options," including retaining the University's existing policy of non-recognition, Tilghman said in an e-mail to PAW. One way to ban Greek life, she said, would be to require matriculating students to pledge not to join fraternities or sororities...

  • decompensation (medical n.): a system's eventual failure (after adapting to a disorder). Something Greek about it.

  • cadastral (n.): by taxable value; used exclusively for official maps. Wolfish.

  • usufructuary (n.): the holder of a usufruct, a legal right to gain from another person's property; they have the right to use (usus) the property and enjoy its fruits (fructus). I love this, it's like a five-year-old's version of Latin.

  • empennage (n.): Aeroplane tail assembly.

  • Quack Miranda (US adj.): The stock text that bogus alternative medicine and health foods to duck legal responsibility for their passive espionage against medicine and dietetics, a la the mandatory Miranda rights:
    These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure or prevent any disease.

  • data turking: generating labelled data via a low-pay artificial artificial intelligence platform.

  • WAR file (n.): Web application ARchive. A Java thing. (I came across `server.war` at work.)

  • ATR (corporate abstract n.): authority to recruit

  • strats (corporate adj.): quant development. Originally a Goldman Sachs term, copied fast. Original contraction isn't defined anywhere, but "strategies".

  • quantoid (adj.): pejorative term for 'quantitative' among (innumerate?) sociologists.
    Also 2) The side of a linear differential equation which is faced by a zero.

See also this list of words Nabokov dug up or invented.


great algorithms reference

From an off-hand comment by Gwern:
Fast, simple, general - a good statistical method lets you choose one; a great method lets you choose two.

Fast Simple General
Y* Y
Neural nets

* Simple to interpret results, not to build well.

London on £25 a day

  • £0.02 : earplugs
  • £16: room in a flatshare with 3 others in far East London.
  • £1.50 : power smoothie breakfast
  • £3 : Tesco lunch
  • £1.50 : Peter Special dinner
  • £1 : Misc (toiletries, friends, whatevs)
  • £0.40 : supplements (amortized)
  • £0.40 : bike maintenance (2 x 6 miles per day commute is hard on it) (amortized)
  • £0.30 : charity shop clothes (amortized)
  • £0.20 : internet
  • £0.00 : council tax (included in rent)
  • £0.00 : library books
  • £0.10 : electricity
  • £0.00 : heating. it's London.

This isn't a minimal per diem - you could live in a car, or benefit from nepotism of some sort - but pretty optimal in 4D nutrition/comfort/employability/thrift space.


slingshot akrasia

Everything on this site was written in the glow and shadow of other things I should have been doing.

This is a further great benefit of work, formal study, and love alike: they pressurise my life. They give me a structure to defy and be inspired by, a gravity assist. I am happiest when laden with obligations, when the set of tasks that is my life flies just out of control, when deadlines tighten. I haven't crunched the data yet (that is, modelled my output vs my obligations) but I am 80% confident that taking on more improves mood and productivity, up to some threshold I haven't found yet.

(To give this vague grandiosity some substance: I'm currently working full-time in a technical field that is new to me, finishing a part-time maths degree, in an intense long-distance relationship, working on four or five software side projects, completing two longish MOOC specialisations, and reading three books.)


I often wonder what kind of person I would be if I had been protected from the cold wind of fate by the screen of wealth... to reach the tawdry heights of being a good assistant book-keeper in a job that is about as demanding as an afternoon nap and offers a salary that gives me just enough to live on.

I know that, had that past existed, I would not now be capable of writing these pages, which, though few, I would undoubtedly have only day-dreamed, given more comfortable circumstances. For banality is a form of intelligence, and reality, especially if it is brutish and rough, forms a natural complement to the soul. Much of what I feel and think I owe to my work as a book-keeper since the former exists as a negation of and flight from the latter.
– Fernando Pessoa

It is just his pipe dream, a vulgar folly he retains simply to prove to himself that men are still men and not the keys of a piano; it is a folly threatened so completely by these laws of nature, that soon one will be able to desire nothing but by the calendar. And that is not all: even if man really were nothing but a piano-key, even if this were proved to him by natural science and mathematics, even then one would not become reasonable, but would purposely do something perverse out of simple ingratitude, simply to gain one's point...

I believe in it, I answer for it, for the whole work of man really seems to consist in nothing but proving to himself every minute that he is a man and not a piano-key!
– Dostoevsky

But the struggle against Plato -- the struggle against the ecclesiastical oppression of millenniums of Christianity... produced in Europe a magnificent tension of soul, such as had not existed anywhere previously; with such a tensely strained bow one can now aim at the furthest goals... we, who are neither Jesuits, nor democrats, nor even sufficiently Germans, we good Europeans, and free, very free spirits -- we have it still, all the distress of spirit and all the tension of its bow! And perhaps also the arrow, the duty, and, who knows? The goal to aim at...
– Nietzsche

I have papers to grade, a grant proposal to review, drafts of dissertations to read. I am working on this essay as a way of not doing all of those things. This is the essence of what I call structured procrastination...

All procrastinators put off things they have to do. Structured procrastination is the art of making this bad trait work for you. The key idea is that procrastinating does not mean doing absolutely nothing. Procrastinators seldom do absolutely nothing; they do marginally useful things, such as gardening or sharpening pencils or making a diagram of how they will reorganize their files when they find the time. Why does the procrastinator do these things? Because accomplishing these tasks is a way of not doing something more important.

If all the procrastinator had left to do was to sharpen some pencils, no force on earth could get him to do it. However, the procrastinator can be motivated to do difficult, timely, and important tasks, as long as these tasks are a way of not doing something more important...

Doing those tasks becomes a way of not doing the things higher on the list. With this sort of appropriate task structure, you can become a useful citizen. Indeed, the procrastinator can even acquire, as I have, a reputation for getting a lot done.

Procrastinators often follow exactly the wrong tack. They try to minimize their commitments, assuming that if they have only a few things to do, they will quit procrastinating and get them done. But this approach ignores the basic nature of the procrastinator and destroys his most important source of motivation. The few tasks on his list will be, by definition, the most important. And the only way to avoid doing them will be to do nothing. This is the way to become a couch potato, not an effective human being...

The second step in the art of structured procrastination is to pick the right sorts of projects for the top of the list. The ideal projects have two characteristics -- they seem to have clear deadlines (but really don't), and they seem awfully important (but really aren't). Luckily, life abounds with such tasks. At universities, the vast majority of tasks fall into those two categories, and I'm sure the same is true for most other institutions...

At this point, the observant reader may feel that structured procrastination requires a certain amount of self-deception, since one is, in effect, constantly perpetrating a pyramid scheme on oneself. Exactly... what could be more noble than using one character flaw to offset the effects of another?
– John Perry

The best circumstance for writing, I realized... were those in which the world was constantly knocking at your door; in such circumstances, the work you were engaged in generated a kind of pressure, a force to keep the world at bay. Whereas here, on Alonissos, there was nothing to keep at bay, there was no incentive to generate any pressure within the work, and so the surrounding emptiness invaded and dissipated, overwhelmed you with inertia. All you could do was look at the sea and the sky and after a couple of days you could scarcely be bothered to do that.
– Geoff Dyer

[After months of doing only my main goal] I took on a job doing closed captioning because I found it [made for] an easier time writing. Just something about talking to people and watching weird media made the writing a lot easier. My new theory of self was that you can't write well unless you have a little strife in your life. I worked at the closed captioning job for 4-6 months and by then I was making enough money on the site to responsibly quit my job.

The problem was I didn't want to quit my job and have readership fall off because I couldn't write, so my crazy idea was to go back to school. I thought, it'd to be this weird environment, with younger people, and that would be good. At some point I switched over to physics because I thought it was really neat, and the comics improved and got more geeky and were a higher quality.
– Zach Weiner

it is not unless I have a formal obligation to defy that I create anything. Worked out a mechanism for why; call it the pinctadan itch:

1. I am fundamentally childish and require a steady stream of variety. 2. Having a job regularises my week: without extra effort, all days resemble each other. 3. Intolerable resentment ensues. I am forced to produce sparks to satisfy my basic drives.

What can you do? You can vary your surroundings or you can vary the furnishings of your mind. In fact three of the most common broad ways of living divide right down this line - bohemianism (artists, students, hipsters), 'grown-up' professionalism, and nerd culture (which straddles the line).

Work precludes variety in your external surroundings from day to day; so you have to
internalise variety. Bohemian life precludes all sorts of things, but it does let you sample any part of reality which does not require any money or power (insofar as your Couchsurfing and Workaway rep is good)...
– me a while back

I don't know if this is ridiculous or platitudinous: I really think this "slingshot akrasia" (structured procrastination) is a central fact of my psychology. (It is somehow related to how great I feel when I don't have to go to a party, to my sadly efficient approach to my grades, to how giving work to a busy person is a good way of getting it done quicker, i.e. an implausible linear increase of output with increasing things to do.)

It is possible that the grand narration above is delusional, and that the only actual content here is "A lot of people work better under pressure".


entrée Noûs

  1. The brain is constructed entirely of ingested matter.
  2. Knowledge inheres in the brain.*
  3. So knowledge inheres in (metabolised) food.
  4. Food, like all matter, is noumenal, of the external world.
  5. So the mind inheres in the external world.
  6. So there is no metaphysical barrier between mind and world.**
  7. So there is no high-level puzzle about knowledge.***

* Yes, not just in the brain, but this suffices.

** Clearly this does not defeat the radical sceptic in her original, Cartesian internalist problematic. But the best candidate for a philosophical fact is: nothing can. Their simple, hard-reset reply is just: "it's an epistemic barrier, not a metaphysical one".

*** All this leaves to solve are the smaller titanic mysteries of consciousness, phenomenal binding, meaning, apriority, most of the highly unfinished fields neuroscience, cognitive science, behavioural genetics ...


notable denoters

  • Advanced Persistent Threat (infosec n.): In cybersecurity, the worst foe. These people are 'Advanced' relative to a script kiddie or a skilled troll. Better funded, more patient, able and willing to try several different avenues of attack. A state actor (or a corporate black operator in a cyberpunk book). They will get in: the question is if you notice, and how much damage they do when they do.
    A conventional hacker or criminal isn't interested in any particular target. He wants a thousand credit card numbers for fraud, or to break into an account and turn it into a zombie, or whatever. Security against this sort of attacker is relative; as long as you're more secure than almost everyone else, the attackers will go after other people, not you. An APT is different; it's an attacker who -- for whatever reason -- wants to attack you. Against this sort of attacker, the absolute level of your security is what's important. It doesn't matter how secure you are compared to your peers; all that matters is whether you're secure enough to keep him out.

  • polycule (n.): polyamorous molecule. Just a cute word for any poly relationship structure.

  • metamour (n.): Your partner's partner.

  • permtractor (UK n.): permanent contractor: someone who works at a company for an extended period of time without being an employee. For tax reasons, or avoiding granting employee rights. The Revenoo apparently classes it as "disguised employment".

  • SLOP (n.): self-selected listener opinion poll. Perhaps the lowest grade of survey evidence, but rhetorically powerful because we are idiots.

  • PTO (corporate n.): Paid Time Off; HR system which does not distinguish holidays, sick leave, and "personal days". Sounds good, but obvious perversities crop up one uniform cap applied to all: if you get ill after a long holiday, you have to work through it?

  • cotton ceiling (n.): putative exclusion of trans women from high status places in women's lib. Apparently has been presented very foolishly: the original post seemed to shame lesbians for not having sex with trans women. (But, if we had to throw out every concept that was ever used stupidly...)

  • force de frappe (Fr. n.): Strike force: the French nuclear deterrent. Once involved serious land launch capability: but who would want to be next to a Pluton or a Hades?

  • acheteur (Fr n.): Buyer. But it could be anything, just so long as it was grand, high in gravitas and dignite.

  • sqeuclidean (adj. or n.): The squared Euclidean (space). Easier to implement on computers.


data science

If you think academic social science is bad, you should see what goes on in corporations.


Done in February 2017

  • Sat an IQ test. Pride: 1, irrationally *
  • Maths assignment #5, Jacobians and Fourier analysis. Pride: 3.
  • Started making SymPy uni notes. Pride: 4
  • Big idea: an Age of Em visualisation. Pride:
  • Wrote about learning but not internalising. Pride: 3.
  • Wrote a snarky little bit about the qualitative and quantitative. Pride: 1
  • Wrote about Pi and Tau. Pride: 2.
  • Wrote again about the maximum wage. Pride: 1
  • Fixed the pipeline (not my job) with bash and long hours. Pride: 3.

* I once argued with a gay friend about what I saw as misuse of the concept "pride" in world Pride events. You should only be proud of things you have actually done, not just proud of who you are.** I insisted that the broader concept they're looking for - in conducting an intentional, public valuation to counter ancient and systematic degradation - is "esteem".

In the general population, I argued, this confusion is one cause of the vast and hollow parade of achievement-free self-promotion and celebrity, as evidenced by e.g. Instagram. I opined that there was no reason to think that LGBT people would escape this corrosion if the distinction between earned pride and unearned esteem were not popularly upheld.

I further offered three resolutions to the contradiction of a Pride march:
  1. Make it clearer that the pride is due to being out, still a very brave action in most of the world.
    Alternatively, proud of having survived a homophobic upbringing.
  2. Reject the very strong evidence of a genetic influence on orientation, and suggest that all instances of nonheterosexuality are chosen.
  3. Change it to "Esteem".

She replied, "Nah, we'll just alter the usage of 'pride', cos 'esteem' doesn't sound as good".
And I said, "Och, ok."

** Sigh, yes, except insofar as you actually have invented your present self, which is a non-negligible amount for any functioning adult.

notable Worten sind Taten

  • nut 'graph or nut graf (n.): That ugly, mise-en-scene opening paragraph in every news article ever. Tries to answer who, what, why, when, where in two sentences.

  • vanity shingle (US pej. n.): a small film production company founded by a celebrity for projects starring themselves. Incorporates shingle: archaic metonym for a small company, via the chalked signboard showing its name outside; or the set clapperboard?

  • bomb-ass: exemplary; highly laudable; successful, op. cit.

  • gaggle (US n.): an off-camera press conference with the White House Spokesman. Fairly routine, but recently made sinister by selectivity.

  • to calque (v.): to translate word for word, or to give an etymology, instead of actual contextual usage. e.g. this clumsy note from Arrival.

  • garnishment (US n.): an ongoing, court-ordered deduction from your wages. To pay fines or debts. Something about the sound is sinister.

  • erfi (n): the imaginary error function. One day I might stop encountering conventional maths symbols I have never seen before.

  • overdimensionality (n.): property of a dataset: having an excessive number of features given available computations. A tight shibboleth for data scientist or Gnostic Muslim.

  • to handfast (Anglo-Saxon v.): to pledge; to contractually sign.

  • handfasting (n.): 1) C15th Scotland: a marriage, not necessarily permanent by design; 2) appropriations of the latter by hippies (e.g. Jim Morrison).


Things I would do differently: Education

I am a remarkably unreflective person. I go months at a time without thinking about my past, or the people I once knew. And not because my past's fucked up or anything; just because the present and the far, far future crowd it out. I'm pretty happy with this arrangement.

Recently, though, I've realised some easy things I could have done to be a better writer / scholar / researcher as of 2017. (They are hardly tragedies though, just inefficiencies.)

1. Physics

Picking courses as a 17 year old in a country without tuition fees, I latched on to the most obvious sources of meaning: philosophy, music, literature. But I could have gotten into physics or stats or computer science if I'd applied (I did get in for biology). And these would serve my present purposes much more, because I'm aiming at truth, and these latter are our greatest machineries of truth.

I don't regret my MA. (Though I probably would if I were English.) Formal philosophical study - that is, seeing what knots and messes the greatest minds in history have tied themselves into, working off no data - has probably saved me from some errors people make when they slip into metaphysics unawares.

And it has probably made me less overconfident that the world can be solved by pure, solitary thought. ("The penalty for not doing philosophy is giving bad philosophical arguments a free pass.") And I have a thick layer of protective scepticism about macroeconomics.

But I would have read philosophy and poetry anyway - I have a great appetite for them, and had it before I got institutional grounding - and so would have gotten much of the inoculation against bad philosophy and the realisation of the relative shallowness of great artists even had I done something harder.

As it is, I've been scrabbling to piece together an education in scientific modelling ever since graduating, and it has taken ages on my own, and I am quite sure that I did this backwards. (Needless to say, the average 2010 economics curriculum was not scientific enough to count.)

But ooh. "Inoculation against bad philosophy and bad economics": is this is the most positive case I can make for my classes? Yes but never mind classes: the greater part of the value came from having 4 years to straighten out my head, and a hundred wonderful people from over the world to collide with, brighter than anyone I'd known before. But again, I'd probably have found them as a physics boy; it was a small university, and my nature is not so malleable.

The distinctive value of an arts degree - that it draws creative misfitting people, that it's low-intensity enough for you to have many projects and loves without constant stress, that it permanently demystifies the baroque, ridiculous world of high culture - are wonderful, but I think I'd rather know how the world works, on balance.

2. Code everything

After my arts degree I switched into software development, a viscerally satisfying career to me. Not just talk, not just interpretation: but fucking building things.

But as well as a fun career, code is an incredible way of expressing thought. You get an oracle, the compiler, tell you if it could possibly be true.

See, coding is a novel way of thinking in general. Yes, it is like maths - but testable, causal, interactive.

A programming language is "how you tell a computer what to do". But before that it's a way to express ideas and get push back from a rational oracle. (It's not reality that's pushing back, of course. You don't know if they're true, but you know if they are clear, if they could even possibly be true, if you are not completely fooling yourself.)

Consider the Bible, or Karl Marx's work, or Sigmund Freud's work. These are rammed full of invalid and unsound ideas - but they are beautiful, unified, and powerful, so they proved persuasive to billions of people. Human language offers no easy test of consistency, no way of really precisely connecting idea to idea. We have had only hard, piecemeal, irreplicable interpretation.

To see what's added by code, here's a thought experiment: Imagine the economic value of a line-by-line description, in English, of the Linux kernel. It would be nothing compared to the billions of dollars of value the kernel has created or saved.

The computability of source code is a side effect of its clarity. Code is testable thought.

I'm converting my maths notes into Python as a matter of urgency, because standard Mathscript is not good. I don't know why this took 2 years to occur to me; clearly the claws of the arts run deep.

This macroeconomics course, in Python and Julia, has crystallised a host of things I only mechanically learned before.

In philosophy, it would have let me get into the thriving and objectively progressive research programmes: philosophy of information, logics, cellular automata, and so on. Here are two great examples of coded philosophy, as proof of concept.

3. Use the blogospheres.

I have learned more about economics from reading Hanson, Quiggin, Krugman, Caplan, Dillow, Friedman, deLong, Harford, Cowen, Sumner, and Smith, than I did in two full years' worth of lectures at Aberdeen. Which is strange, because most of them are academics. But, because their readers are from broad backgrounds, the writing is vastly superior to that of papers: clearer, briefer, and more easily evaluated for both rigour and well-foundedness. In 2010 the econ 'sphere wasn't as highly developed as it is now, but was still good enough.

In stats, Andrew Gelman, Uri Simonsohn, and Cosma Shalizi's blogs have taught me what's wrong with science and how to fix it, which I didn't get a jot of in classes.

(Philosophy and maths benefit less from this, because their usual texts are more digestible and more ineliminably systematic, respectively.)

This step wouldn't have improved my grades much, because of teaching-to-the-test.* But it would give me what universities are supposed to give: firm grounding in expert knowledge about things which matter, and the ability to apply it appropriately.

* A dark implication: that one could be better-off, in finances but also in knowledge, without uni altogether. (Since they distract you with password learning and rote crap.) We rely on the spiritual and psychological gains of 4 years of relative leisure. And at the micro level, this is a clear good deal.

4. Focus

Over the past 4 years, out of uni, I've read an average 102 books a year. They have been about everything, and it has been wonderful. A four-year cruise on about £300.

But I am persuaded that this isn't how you contribute to human knowledge. The absurdities of siloed scholarship - economists and anthropologists and sociologists and psychologists and all talking about the same thing, but wholly ignorant of each others' insights - are large, and can't be fixed except by people who own several hats. But everything else is done by specialists, because the coalfaces of knowledge are very far from common sense, in several different directions, and anyone who tries to reach several of them is likely to end up near where they started.

One of my resolutions this year is to read fewer than 25 books, but to make them all count. I have a folder, "Spoilers for Reality", with textbooks and serious crap to get through. (In each of those hundred-book years I was supposed to be studying maths, and you can imagine how much I actually did.)


notable mental methane vents

  • SOC (n.): system on a chip. Previously known as a microcontroller. What we now call just 'a computer', but integrated boards were an enormous deal, a revolution within the digital revolution.

  • UUOC (n.): Useless Use Of Cat (Award). Surprisingly mean retort to StackOverflow answers which use the UNIX tool cat where a pipe would do.

  • abience (n.): the urge to withdraw. Usually used to mean pathological avoidance, but to me it is also the plain, sacred joy of missing out.

  • hardtack (n): A very basic cracker, just baked flour and water. Staple of navies and Tudor explorers.

  • HARKing (v.): Hypothesizing After the Results are Known. A particular problem in social science, where pre-registration of studies is a tiny minority of work.

  • merchantable (UK legal n.): Good enough to be sold.

  • technical steer (n.): Input from expert staff, AKA 'knowledge'.

  • whitespace damage (n.): subtle but breaking changes to source code performed by ordinary text processors, e.g. line wrapping, hidden characters, odd apostrophes. This phrase is a shibboleth for being A Very Serious Person, e.g. a kernel dev.

  • moving up the value chain (phrase): performing work further away from physical extraction, processing, and manufacturing. Supposedly insulates you from competition because your outputs are less easily evaluated as they become less physical. Economic abstraction.

  • FANG: Facebook / Amazon / Netflix / Google, particuarly when their stocks are used as a bellwether.

  • DGP: data generating process. This took bloody ages to google.

  • rebranding (v.): "a euphemism for 'euphemism'" - Jonathan Meades

machines inside

PSA: It took me many years to internalise the formal methods I know now.*

I use "internalise" as distinct from "learn", because, let's face it: we all "learn" statistics in uni, in the sense of briefly knowing a tiny set of teacher passwords, of knowing what a mode is, and of knowing how to dumbly apply two canned tests of inference.** But almost no-one with that badge on their resume actually remembers, actually uses, and was actually changed by contact with it, the driest and most nutritious method.

My measure of internalisation is if you use the method, without prompting by school or advisor, in your investigations. Internalisation requires some understanding, but I'm not saying that I have any deep grasp of these things. I just appreciate their power, and use them as well as I can where I can.

  • First contact with algebra: 2000
    Internalised algebra: 2012.

  • First contact with Analysis: 2003
    Internalised Analysis: Not yet.

  • First contact with formal logic: 2008.
    Internalised first-order logic: 2010. (pic above)

  • First contact with proof: 2010
    Internalised proof: Not yet.

  • First contact with statistical inference: 2011 **
    Internalised statistical inference: 2017.

  • First contact with Bayesian / cognitive / decision science: 2010
    Internalised decision science: Not yet.

  • First contact with full-blown probability theory***: 2012
    Internalised probability theory: 2017.

  • First contact with (imperative) computational thinking: 2014
    Internalised computations: 2015.

  • First contact with functional programming: 2016
    Internalised computations: Not yet.

  • First contact with machine learning: 2015
    Internalised ML: Not yet.

* This strikes me as worth stating, because the rigorous fields are so demoralising to tackle alone, and take so long for even very intelligent people to get comfortable with. In a standard mathematical education, we don't get to see the cockups or the thousands of fruitless hours that Jacobi or Germain had to put in, to win as they won. (The painstaking labours of Wiles and Zhang are at least a bit more available to us.)

** I have tried to learn statistics (that is, higher statistics, data analysis and inference) four times in my short life:

  • 1) 2010: in the standard, cursory Research Methods module in undergraduate economics (I find myself not guilty).
  • 2) 2011: to catch up with discussions on LessWrong
  • 3) 2013: through a formal stats degree
  • 4) 2016: on the job.

Only the last is sticking.^ But I was extremely lucky to get a statistically demanding job without credentials in the first place; I snuck in because my profession confuses people and my programming ability was so far beyond the spec that they halo-effected me in.

So where does everyone else actually learn stats?

^ Again, I'm using a strict definition of "learn": in the others I learned many terms and followed many formal derivations and clicked many buttons in many statistical packages, but I did not actually gain the statistical mindset, modelling ability.

*** Not even measure theory...