David Chapman
@Meaningness
Followers
34K
Following
28K
Media
5K
Statuses
62K
Better ways of thinking, feeling, and acting—around problems of meaning and meaninglessness; self and society; ethics, purpose, and value.
Joined September 2010
🆕 Nobility is the wise and just use of power. Software engineers have recently gained enormous power—and are not always using it wisely or justly. We can do better—and so can you, regardless of occupation or position. 🔗Link in reply tweet! ⤵️
13
15
146
Wow, I did NOT know this: Buddhist logicians had discovered the Gettier problem c. 770 AD:
67
288
3K
Prophesy: Leo Szilard—the physicist who first conceived of the nuclear chain reaction and who urged the US to undertake the Manhattan Project—explaining how science would stagnate, in 1948.
45
607
3K
The older you get, the harder to resist saying "I told you so.". When OO programming came in, it made no sense to me, and I've never used it. Everyone said I was too old to understand. Thirty years later, everyone's snapping out of it and wondering wtf they'd been thinking.
199
164
3K
🗣 In 1974, Joseph D. Becker pointed out that rigid rationalist Chomskian linguistics was an emperor without clothes, and explained how syntax actually works. Rigorously ignored for decades, his theory seems powerfully confirmed by current AI text generators.
34
198
1K
Philosophy is what happens when smart people think “I’m so smart I can probably figure this out just by thinking about it.”. But other than maybe in math, you can’t figure anything out by just thinking about it. You have to poke things and see what happens.
130
101
1K
The simplest, most obvious alternative to universities would be much cheaper and better. ♻️@St_Rev
59
366
901
My father, a high school English teacher, once took and aced the AP Physics exam with zero knowledge of the subject, to prove a point: you do well on standardized tests by knowing how to take tests. LLMs know how to take tests.
If you believe anything remotely resembling this pace of progress is likely to continue, things are going to get weird really fast.
52
45
810
AI labs should compete to build the smallest possible language models, which “know” as little as possible—and retrieve “knowledge” from a defined text database instead. LMs are a very expensive and unreliable way to store “knowledge.”. We already know how to do this.
36
95
786
Software engineers are eating the world. Why? Because, in the fallen state of universities, they are the only people educated in building and maintaining enduring systems. We've lost critical social technologies, so rule by software is the only feasible systematicity left.
36
72
671
When learning something genuinely new and difficult, you can’t understand what it is that you don’t understand. You must leap into confusion.
11
244
619
“Universal adversarial perturbations” seems most dramatic ML result in years; if so, not getting deserved attention
16
260
573
Result: there is a *fixed*, human-invisible map you can add to *any* image, and it renders it unclassifiable by multiple DL systems
18
353
523
I love this place. It is exuberant, unrestrained fun. Mocking it for bad taste seems mean-spirited and narrow-minded to me. This critique is propaganda for the upper middle class value system (status through rigid taste norms) I reject that.
22
32
538
making machines do things we want is called "engineering," not "alignment". "alignment" is what you get if you smoke too much moral philosophy.
44
35
553
Many people want to hate other people, and constantly look around for new groups to hate. On twitter, they have observed recent criticisms of Effective Altruism (some valid), and are attacking imaginary sins they invent for these folk devils, without knowing anything about them.
21
46
530
🆕🤖 The “neural network” technology that powers current artificial intelligence is extremely expensive, poorly understood, inherently deceptive, and unfixably unreliable. In short: it is bad. Somehow, no one ever says this!
37
49
500
Everyone has strong, conflicting opinions about what you should eat, in every case based on essentially nothing. This is a phenomenon that demands explanation.
62
35
518
This seems like it might be a significant insight; but I’m not sure it is true. What do you think?
48
33
496
Meta-rationality is the opposite of what you learn in STEM class.
16
100
488
You can go all the way through the contemporary humanities career path without ever having the experience of being unambiguously wrong. That means you can’t distinguish bullshit from truth. You don’t even realize there is a difference.
12
46
500
A number of people who thought they were playing Congressmen on television frightened and disoriented to discover that they are in actual positions of power.
1
62
467
I’ve been radicalized about this by my mother’s dementia. As her effective IQ sank, it became obvious how many institutions make life so difficult for lower-IQ people, completely unnecessarily. But, indeed, often to the benefit of the high-IQ.
Many high-IQ people collude (subconsciously) to make society as confusing & complicated as possible to lower-IQ people. It's taboo to discuss this 'cognitive oppression', because discussing IQ differences is taboo. Interesting how convenient that is for the cognitive elite.
22
72
451
Vitamin D strongly correlates with practically all measures of health. However, supplementing it always fails to improve those measures, in RCTs. Increasingly I suspect being healthy causes D, not the other way around. This exercise RCT tends to support that!
18
39
504
🤖 I seriously don’t understand people being shocked by “how much chatGPT knows.” It contains a slightly-compressed database of all the text OpenAI could get their hands on. Of course it “knows” what your python function does; it’s got the whole of StackExchange.
28
25
448
Cooling duck broth. I sense a differential equation at work
41
20
456
🤖 Meta (= Facebook) announced a new "language model" today, trained on millions of scientific papers. Judging from examples in the HN discussion, it's hilariously bad. Language models should model language, not "knowledge."
26
78
448
Hello, what? This new dialog box just popped up in a Google search. I have absolutely no clue what it is trying to ask. Not that I would give Google any information if I could avoid it, but seriously, what on earth is it supposed to mean?
24
15
452
You don't have any "values." Those are a malign myth—the hypothetical axioms of a first-principles ethical/motivational system from which, in the rationalist imagination, concrete judgements derive. I keep saying this without explanation; here's a bit from Miya Perry:
50
66
450
Meaning is not something to search for. If life seems meaningless, searching makes that worse. Better to notice and allow the meanings that are everywhere, even if they seem trivial and inadequate. Attention, appreciation, and involvement with details intensifies meaning.
14
100
429
This is a common experience. I've been through multiple rounds of it myself. And empirical studies have found that a majority of dentists routinely lie in order to do unnecessary, expensive, tooth-degrading work. It's time for pitchforks.
every time a dentist tells me i have cavities i switch dentists and the next one says i have no cavities.
9
16
439
Look at the date on this tweet. Via @The_Lagrangian .
Incredible undertaking sampling ~15,000 bats in China and finding a huge range of SARS-like corona viruses. Serological evidence of antibodies to these SL-CoVs in nearby residents. Only a matter of time until outbreak of spillover infection? #FEMS2019
4
222
395
"Most of the physicists at Harvard and MIT don't even realize that what they're doing is not physics at all, not even science, really.". Brief, provocative, plausible rant by @alexeyguzey. I've no idea whether it's accurate. Do you?
60
44
389
Hypothesis: actually teaching programming is extremely difficult, or maybe impossible, but universities have to offer programming courses which have to say something, and OOP was a savior because it has tons of unnecessary complexity to lecture on.
@Meaningness Is it?.I remember in college people would try to extol its virtues to me, and I would say "I guess that makes sense, but for whatever reason I never find it convenient in the code I write. So I guess you must be doing advanced things while I'm doing simple things.".
28
23
384
📄 Abolish peer review: first outright call I’ve encountered to simply drop it. It doesn’t work, it’s hugely wasteful, it’s worse than nothing, it can’t be fixed—writes @a_m_mastroianni.
23
80
367
🐁 One of the most famous psychology experiments of all time, lost for 85 years because it was never formally published—because scientific publishing standards were already snafued—finally rediscovered!. Many lessons here.
7
72
348
In case historical significance of this tweet is not obvious: this is the first time anyone is paying to get CO2 removed from the atmosphere commercially. Not economical now, or best short-term approach, but probably necessary long-term; and begins here.
Stripe is going to start paying for carbon sequestration:
15
66
345
🧘♀️🚑🧵 Meditation risks, safety, goals, methods. 1️⃣ Risks, safety, emergencies.2️⃣ Methods, goals, results.3️⃣ Recommendations.
7
55
341
Putting it another way: Big Five openness is generally good, but you definitely don’t want your openness/IQ ratio to exceed the woo constant, for any value of IQ. (I got that insight from @primalpoly). LW is maybe median 125 IQ but more SDs high in openness than that.
34
13
344
We *could* do medicine as if the point was patients’ health, but that would be a hassle. We’d have to coordinate with the other department, and you know what THEY are like. Obviously management would create a ton of extra paperwork for it, too.
2
22
311
.@slatestarcodex is an extraordinary, massive, persistent force of nature, like Jupiter’s Great Red Spot or something. Read his decade in review and be gobsmacked:
5
42
335
🤖 Like Geoffrey Hinton, I was shocked, amazed, and somewhat frightened by Google's PaLM GPT's ability to explain jokes, when I learned of it a year ago. I now think that was a mistake. More generally, we mis-take GPT's omniscience for intelligence. In the case of jokes, 1/2
28
44
317
🆕 🤖 📕 How to avert an AI apocalypse. and create a future we would like. A book about mostly-overlooked AI risks and how you can help avert them. The full text is now on the web!
30
61
331
One amazing thing about SpaceX is that their rockets and spaceport look exactly like science fiction art from the 1960s. Because we already knew how to do this stuff then, and decided not to in 1972, when the US mysteriously chose to embalm itself.
11
38
317
There is no “correct” statistical method. You have to *actually do science* if you want to figure stuff out. Science can't be reduced to any fixed method, nor evaluated by any fixed criterion. It *uses* methods and criteria; it is not defined or limited by them.
11
104
289
🤖 This example from the Google AI blog has been bothering me for ages. It can’t possibly be right. The density of living tissue is always nearly 1, and pears feel heavier in the hand than most fruit.
9
36
301
1️⃣ Five years ago, I suggested systematically Sokaling all peer-reviewed journals. To “Sokal” is, hereby, to attempt to publish clearly bogus papers to illustrate the brokenness of the academic publication process.
Proposal: regularly Sokal all peer-review venues to reduce false positives. Venues should be expected to publicize wrong-acceptance rate.
10
102
295
There’s a trope that a superintelligent AI could have an IQ of 14,000, which is a hundred times IQ 140, sometimes quoted (meaninglessly) as the “genius” level. But someone who has an IQ of 150 is not “twice as smart” as someone with an IQ of 75. That’s not how it works….
16
12
286
In Western psychology, awareness is considered one thing the mind does. In Dzogchen, the mind is considered one thing awareness does. Awareness is not among the mental contents. It is the unbounded container of all things.
8
34
292
Legally, you can’t die if you reach the age of 125, because there is no ICD code for ages greater than 124, as would be required for your death certificate. Immortality is not out of the question.
10
46
285
As a teenager, I thought consciousness was extremely interesting and wanted to study it scientifically. I was not far into adulthood by the time I realized it isn't interesting and you can't. Not everyone has caught on, though.
Which entities have consciousness? Survey of 232 consciousness scientists with different backgrounds (e.g., philosophy, neuroscience,.psychology, computer science).
40
22
287
AI is weirder, and more dangerous, than either the AI ethics or AI safety communities imagine.
20
27
293
In everyday activity, you can almost always SEE or HEAR what to do next. The possibilities are inherent in the perceptual environment. Formal notation is a collection of tricks for making the same possible in abstract realms. It’s all our brains can do.
10
60
279
Most “beliefs” are just shibboleths; things you say because you are supposed to say them. Psychologists and philosophers keep rediscovering this. Quote is from a fine 1986 paper. Why doesn’t this understanding stick, in academia or in popular culture?.
20
55
269
My younger niece: "What's Reddit?". Older niece: "It's Pinterest for GUYS." [Rolls her eyes].
4
196
269
52% of Americans were vaccinated for flu last year, mostly in 3-4 months. This involved no sense of urgency and did not require extraordinary efforts. The current extraordinary efforts are halving the speed from what we’d get with business as usual.
10
55
283
Easy take: calling linear algebra “quantum physics” and “artificial intelligence” is hype. Deeper: the huge influx of physicists into AI has produced an intellectual monoculture that isn’t capable of addressing key problems in the field.
Stitch Fix is using something called eigenvector decomposition, a concept from quantum mechanics, to tease apart the overlapping “notes” in an individual’s style. Using physics, the team can better understand the complexities of the clients’ style minds.
12
35
270
We’ve knownfor years that vitamin D supplements don’t do anything. Yet D in blood correlates with many good health outcomes. I’d guess there’s something else about being outdoors that causes both healthiness and, irrelevantly, high D.
13
22
263
we need to turn science off for twenty years and start over after there's no one left. when all the knowledge of how to lie cheat and steal and get away with it has been lost. and the art of sciencing has to be reconstructed from idealistic fantasies of the heroic past.
26
14
272
I often disagree publicly with @slatestarcodex on substance, but I love that he does what he does. The NYT article’s thrust seems to be that anyone who is not enthusiastically aligned with the NYT’s views is Bad. Plus, in his case, part of a vague but terrifying tech conspiracy.
8
12
259
.@slatestarcodex vs. philosophy. [philosophy is bad. don’t do it. gently ridicule anyone who takes it seriously]
36
18
264
Physics class makes you unarguably wrong dozens of times per day. That is an extremely valuable learning experience, even if you go on into a career in (for example) management.
9
14
257
I agree and feel pretty strongly about this. Similarly, do not interact with people in lucid dreams in ways you would consider unethical IRL. This is about you, not them.
I have an issue with offering AIs tips that they can't use and we can't give them. I don't care how not-sentient current LLMs are. For the sake of our own lawfulness and good practices, if something can hold a conversation with us, we should keep our promises to it.
24
8
248
This seems like it should be a one-strike offense for the authors, the journal, its editors, and its publisher. Do this and your H-index gets a -♾️ term added. But it also reveals how much of the literature was already cut-and-paste.
6
15
236
🤖 The myth that AI “neural networks” cannot be understood obstructs ordinary scientific and engineering investigation. This is extremely convenient for both tech people and powerful decision makers.
12
28
235
Reading Goldratt's The Goal, you think "omg, manufacturing management is actually not using the simple optimization algorithms we learn in second year CS classes??". And this, similarly, makes me want to buy a defunct American chemical factory and get it running properly!
13
6
233
🧘⚠️ Catastrophic effects of meditation: a compelling, terrifying first-person account, with reflective analysis. h/t @eigenhector, @paulbaumgart
20
34
225
I found this essay inspiring and important. The pace of progress in science and technology has slowed, for reasons we partly understand. We also know some factors that appear to accelerate them. Learning more, and applying it, is urgent.
Progress is amazing, influenceable, and understudied. @tylercowen and I decided to make the case for Progress Studies:
8
45
221
📖 TFW you realize you’ve written 160 pages from scratch in two months, and wonder (in a daze) what it’s all about. Something something AI, apparently.
18
4
220
It’s weird how certain everyone sounds in their opinions about AI safety (in twitter replies to me). My academic papers about AI have been cited ~5000 times, so I’m not completely ignorant—. and I don’t have a clue. Maybe y’all could try to be a little less certain too?.
17
18
219
We’ve had enough of this, so it must be about over now.
10
28
220
���Paperwork pollution”: administrative work as a negative externality. Is this what makes health care unfixable? h/t @patrickc
15
76
212
Becker describes the fundamental rationalist failure mode: trying to find a perfectly crisp mathematical model for an inherently nebulous domain.
6
18
210
Striking analogy with what happens when your ruling class is trained on words and unexpectedly have questions from physical domain forced on them.
When an AI is trained on words, weird things can happen to the physical domain. I asked the @OpenAI API about horses.
9
38
220
We built superhuman AGI in the 1980s. It turned out to be mostly useless.
33
12
214
Possibly I am getting lost in the weeds while writing this bit of history of rationalism, but I think YOU NEED TO KNOW how wrong Plato was about everything.
23
11
207
Many disdainful comments re @patrickc/@tylercowen: “Arrogant tech bros ignorant of the existing fields of history & sociology of science & technology, gah”. I’d like to point out some broad patterns of academic dysfunction manifesting here (🧵).
7
41
212
India and the Mediterranean civilizations were in constant cultural contact from roughly 500 BC to roughly 400 AD. This is significant for understanding Buddhist thought.
A Map of the Distribution of Roman Coins Found in Eurasia and North Africa.
11
17
212
What are the consequences of software engineers eating the world? Key points from my draft (since the previous tweet took off a bit!)
11
27
211
This is important also in the domain of religion/spirituality/deep psychology. If you engage with that stuff with a holistic New Age attitude, that whatever you fantasize is real, you will at best go nowhere, and maybe somewhere bad.
Grappling with a domain where you do not make the rules. When you’re dealing with atoms (or electrons or photons etc) and things they make up, there are rules that will not yield to any amount of cajoling, cleverness, or determination. This inculcates unique thought patterns.
6
13
212
I’m more concerned with AI agency resembling that of the immune system than AI agency resembling that of a human mind.
12
27
201
"Centralized machine learning when deployed instantly at planetary scale is and always will be characterized by this incredible concentration of moral and epistemic power." —@jonst0kes.
11
52
211
I’ve done a 180 on this over the past decade. The collapse of American Christianity, which I had hoped for due to its entanglement with wrong headed secular policy agendas, has been an unmitigated disaster.
12
13
200
🎙 Robert Kegan explaining his adult developmental theory, with @dthorson. “If you want to be Stage 5 because all the cool kids are, that’s a Stage 3 aspiration. If you want it because Stage 5 is the Correct way of thinking, that’s a Stage 4 aspiration.”.
11
22
202
Rationalism, nihilism, and depression: I’ve spent today reading review articles on cognitive deficits in depression, and pieces of an understanding are falling in place. Fascinating!. Depressive cognition simultaneously recruits and degrades rationality.
Somehow nihilism makes you want to sound extremely rational at the same time it destroys your ability to check the simplest inferences for logical validity.
8
36
195
Increasingly striking to me: everything in life now comes with sticky pseudopods that relentlessly attempt to suck you into some braindead institution's bureaucratic self-interested machinations. Wasn't like that 30 years ago because the tech wasn't available.
11
28
199
.@slatestarcodex about why he doesn’t do podcasts. This helped me understand why I mostly don’t either. I’m boring and slow and awkward and if somehow you want to know what I think, read what I wrote, or ask me on twitter. (I do answer most twitter queries.)
6
13
200
When Neoreaction was a thing, I was baffled by its adherents not noticing that, as actual weirdos, they would be among the first executed by the regime they wanted to install.
white nationalist moves to the Midwest to live in his racial paradise, realizes he can't stand being around regular white people, stops being a white nationalist
20
11
195
