Witticisms, parodies, pointed observations, japeries, jocularity, and discursive drollery on assorted topics (psychology, statistics, transhumanism, technology, Silk Road, Bitcoin)
created: 30 April 2014; modified: 27 Jul 2017; status: in progress; confidence: possible; importance: 8


  • Often, I must calculate otherwise than I think. That is called diplomacy.
  • Surprising linear models work at all, when they are the worst functional language ever: only operator *, n variables, n assignments, & output=sum.
  • When critiquing a paper, go for the jugular: any part with the words presumably, obviously, past research, or studies show.
  • Tip: try to read the appendixes of research papers as well. Surprisingly often big results, not just caveats & flaws, will be buried in them
  • There’s always someone who can make a perpetual motion machine complicated enough that you, personally, can’t figure it out.
  • Remember, citizens, fulltext is our first line of defense against science journalism!

    • We can sleep soundly because rough men stand ready with blogs willing to do violence to papers on our behalf.
  • When does measurement error matter? If you can’t write X correlates with Y as X correlates with scribbled survey asking about Y without qualm.
  • The use of confidence intervals rather than p-values is a clear improvement; it makes our difficulties vanish like smoke in a fog.
  • p-value testing is pretty weird, but with multiple correction, it gets even weirder: the more you measure & model, the less you know.
  • Power vs error vs sample size: painful tradeoff of statistics. Sins of omission, commission, & risk aversion - painful tradeoff of life?
  • Not doing power analysis with informative priors is searching a haystack of unknown size for a needle which probably isn’t there.
  • Power analysis: if there is enough light to find the needles in the haystack; variance component analysis: if there are any needles at all.
  • It’d be nice if people cared about systematic vs sampling error, Bayesian methods, informative priors & design more than once every 4 years.
  • Remember: p-values are not posterior probabilities are not effect sizes are not utilities are not profits are not decisions.
  • All analyses are ultimately decision analyses. We can ignore that for convenience, but we must never forget that.
  • Our beliefs are continuous, but our actions discrete. This leads to analysis paralysis; can one execute wholeheartedly on +EV actions with P=1% just as well as when P=99%?
  • Disagree and commit to courses of action - Thompson posterior sampling for solving the MDP of life.
  • Beware geeks bearing percents.
  • To mine for forgotten gold in a field, look for its most austere & mathematical subfield - mathematicians are terrible at marketing.
  • We will know deep learning has succeeded not when we anthropomorphize machines, but when we unanthropomorphize humans.


  • Whenever I feel maybe the FDA and big blind clinical trials are unnecessary, I go read Longecity threads until the feeling passes.
  • Paleo? Atkins? Low-carb? Mediterranean? Maybe it is time to step back and ask: what should be the epistemology of food?
  • Every normal man must be tempted now & then to sharpen knives, hoist a black flag, and run amok, shouting No causation without randomization!
  • If you consider them in terms of QALYs lost, school is at least as harmful to your health as smoking, and far more expensive.
  • Paul Krugman notes you can look like an insane raving loon just reciting economics textbook contents in a loud voice; likewise, behavioral genetics.
  • Some ask if behavioral genetics has made me nihilistic and feeling all is futile. But I have to believe in free will - my whole family does.



  • The soul of modern man is so fallen & mutilated that he can feel despair only when trapped in dependency hell.
  • One programmer’s dependency solution is another’s version conflict.
  • Whether to check preconditions before a loop, or after: this is to define sanity and insanity.
  • Programmers’ chairs and keyboards cause RSIs of the body; but what RSIs of the mind?
  • Will a just & merciful developer condemn good programs to bad smartphones?
  • We have abstracted away from for loops over arrays; but what idiom will abstract away the loops of our lives?
  • Backups are confronting one’s fallibility & the transience of the world; we should not be surprised so few can do it.
  • When was the last time you saw Americans worry about Indians? A lesson there. But don’t ask an Indian developer what, ask an American.
  • The pre-Internet PC was the real Wild West: a lone programmer and his compiler against hordes of quandaries.
  • A pause on my keyboard for thought - and how peaceful it must be in the CPU, as the nanoseconds slowly tick by…
  • Bugs in working code are moments for reflection: how much we take on faith, because it seems to work!
  • Helping newbies requires an active memory - of all the times we ourselves failed to read the fucking manual.
  • Is programming Christian or Buddhist? Ask yourself how many of your programs have the Three Marks: anitya, anattā, & dukkha.
  • Once a program has taken on a definite form, it does not lose it until deletion.
  • How can we fear machines will separate humans when even in their source code, we can read the stamp of personality & style?
  • Data, and the formats by which data are communicated, inevitably create a system permeated by illusions.
  • A truck driver may drive for 20 years without ever improving; it must be confessed that many programmers do little better.
  • Anyone satisfied by last year’s code is not learning enough.
  • I do not like this algorithm/language/tool. Why? I am not up to it. - anyone, ever?
  • The Internet is embarrassed by the browser.
  • The HTML page is a stark data format: everywhere it goes, there is duplication of process. It is perfect for hiding information.
  • There is but one constant in every program; and it is [see VM table entry 0x000007FEFC831010 ~> 0x00885ED010 ~> faulting in page...]
  • When you become frustrated with computers, please remember they are only cleverly arranged sand. (When you become frustrated with people…)
  • Uproot your functions from their ground and the dangling roots will be seen. More functions!
  • I think there is a world market for maybe five computers - Google, Amazon, Facebook, Microsoft, and Baidu.
  • The summer garbage - / the sole remnant of many / bright engineers’ dreams
  • Like in security, the worst possible state for a power cord or USB cable is half working.


  • To test whether a language despises its users, merely see whether if (a = 1) {...} is valid.
  • Web browser developers are condemned to reinvent the OS, poorly. What is to be done about this? What could ever have been done about this?
  • Mozilla now has its own community, browser, cloud, language, userland, and in a final concession to the obvious - Firefox OS!
  • C, C#, Go, R, Rust: are these unsearchable names testimony to some feet of clay, or testaments to monstrous egos?
  • Haskellers knows the type of everything & the value of nothing. Unless they’ve turned on extensions, then neither
  • C - fast and efficient and for when you don’t have enough memory to remember things like why you don’t want to use C.
  • C programmers stay sane by imagining that all the other insecure buggy programs are thanks to avoidable unrepresentative reasons.
  • We have made an AI breakthrough! With Prolog, we have created the intelligence of a 2-year-old child: No. No. No. No.
  • Creating a new good programming language is so difficult it tends to be only fools who try.
  • Just as winds preserve seas from stagnation, so also corruption in languages is the result of prolonged calm.
  • That code is ill-written of which one must repent; as long as the PHP bears no evil fruit, the fool thinks it sweet as honey.
  • Single-paradigm languages are admirable for pushing until it breaks - or doesn’t!; cf Einstein & Brownian motion; Columbus & spherical Earth


  • If you find yourself surprised by man or market, remember you have learned as much about your own thinking as them: revelation comes in twos.
  • Bitcoin involves no new primitives or fancy proofs; perhaps cryptographers should all along have been studying sociology, not mathematics.
  • The horror of Bitcoin: money really is a social construct! & not always by nice-smelling well-groomed people in sharp suits.
  • A modest proposal: end tenure for computer security researchers. Given the status quo, if they’re not rich, they can’t be any good.
  • What is the price of two satoshis - one copper coin? But not a single satoshi can fall to fees without your Eternal Blockchain knowing it.
  • The bravery of innovators: what sustained Satoshi during those lonely days in 2009 when no one cared enough even to attack Bitcoin?
  • One weird philosophy trick for analysis! Imagine worlds where X failed: Sure Bitcoin died: a deflationary currency requiring ever more waste?
  • Let us hope Karpeles will not be the Alcibiades of Bitcoin.
  • The optimal number of double-spends is not zero.
  • DPR was a genius until he was a fool & a knave; Satoshi was a genius until…?

Darknet Markets

  • Ulbricht, Ulbricht, Ulbricht! A million coins were not enough for Ulbricht!
  • We criticize Ulbricht for not knowing when to quit & enjoy life; but what are we ourselves refusing to quit?
  • Everything looks permanent until its secret is known. For all too many things, the beginning of fear is the beginning of knowledge.
  • A beginning is the time for taking the most delicate care; this every student of the OP SECurity knows.
  • Modest proposal: let’s abandon the term OPSEC. Instead, let us refer to the mistakes of Ulbricht, Benthall, MDpro, etc as OOPSEC.
  • Is crypto-libertarianism self-undermining? Can we handle the sausage factory of the darknet markets?
  • When a centralized escrow market claims 1% commissions, it should append an asterisk: and a one-time 100% fee if hacked or shut down.
  • Multisig vs centralized escrow: proof that convenience is a drug faster & more addictive than the finest heroin.
  • Decentralized Bitcoin markets are the future of darknet markets - and I fear always will be.
  • Lust for lucre is the root of all evil? Alert the theologians: the number of centralized darknet markets is an index of Satanic activity!



  • How would the world look any different than it does now if X were true?
  • Ricardo’s theorem is the pons asinorum of economics: it always seems to be employed by asses with oddly high salaries.
  • To learn to build sandcastles on the beach is to learn to live and die an atheist.
  • Good results follow good intentions even as the rain follows the plow.
  • Neo, what if I told you… everything you knew was correct? (That would be the biggest disappointment ever.)
  • At times, tolerance can be the most radical of positions to take; just watch when a weak group gains power.
  • Reading a flame war: and all you people must once have been little children, who smiled of a summer day.
  • Imagine a world where the just world and fundamental attribution bias were correct, and karma existed. Wouldn’t that be unspeakably tragic?
  • War seeks to mold physical conditions as one wishes; art seeks to mold minds as one wishes. Both are based on deception.
  • Art is often hatred: it shows the past, or future, or far-away but never the present moment. Anti-art: a TV & camera showing the viewer.
  • The fallacy of gray or the fallacy of grey? My suggestion the fallacy of græy was rejected without, I thought, proper consideration.
  • If pigs were smart enough to worry: the humans neither love nor hate you, but you are made of tasty bacon they can use for something else.
  • On genetic engineering and AI risk, optimist: we can’t afford to worry about risks with survivors; pessimist: we can’t afford worry about risks without survivors.
  • I have made progress in my meditations: yesterday I told myself one truth, and only nine lies.
  • Blackmail is a rent on hypocrisy, either personal or societal.
  • RCTs : traditional Western medicine :: Nixon recordings : presidents :: DNA testing : courts :: priest abuse : Catholic Church :: cellphones/body cameras : police :: Wikileaks : geopolitics :: Snowden : computer security
  • Optimists: everything selected to be as efficient as is possible to achieve; pessimists: everything is selected to be as terrible as is possible to get away with.


  • An aphorism is an algorithm, of we know not what input, we know not what output.
  • It is easier to write an incorrect epigram than understand a correct one.
  • Voltaire’s Third Law: for every aphorism, there is an equal and opposite anti-aphorism
  • The epigram is a compressed, golfed, idea, with all the virtues - and sins - of golfed code.
  • If we measure the entropy of epigrams by how many people understand them, who is the gzip of epigrams? The xz? (…the ZPAQ…?)
  • Remember! Most strings are incompressible, most reals uncomputable, most theorems unprovable, most programs undecidable.
  • Should languages support the writer’s convenience or the reader’s understanding? Frame it as a status debate, and all becomes clear.
  • Fear not known but unknown propaganda; I have the utmost respect for _Pravda_s - how else will you know what to not believe?
  • Writing down your ideas takes 90% of the time; debugging code takes another 90% of the time; rewriting takes the third 90% of the time…
  • A day with nothing unusual is an unusual day. Fiction writers & GANs must include the improbable to be probable. Spend shrewdly surprisal.