Skip to content

Instantly share code, notes, and snippets.

@ivan
Last active November 2, 2024 18:42
Show Gist options
  • Save ivan/a36e2489623469d96c1ad79077b6dcf9 to your computer and use it in GitHub Desktop.
Save ivan/a36e2489623469d96c1ad79077b6dcf9 to your computer and use it in GitHub Desktop.
2024 reading list

Things I might read in 2024.



  • Antoine de Saint-Exupéry, Richard Howard (translator) - The Little Prince
  • (Translation by) Sam Hamill - Yellow River: Three Hundred Poems From the Chinese
  • Sayaka Murata, Ginny Tapley Takemori (translator) - Convenience Store Woman (via)
  • Jorge Luis Borges - Tlön, Uqbar, Orbis Tertius (in Labyrinths)/ printed (via)
  • Franz Kafka - The Metamorphosis (via)
  • William Olaf Stapledon - Star Maker/ audio, go to 12m35s to skip past the introduction spoilers

  • The Heart of Innovation: A Field Guide for Navigating to Authentic Demand/ audio (via)
  • Peter D. Kaufman - Poor Charlie's Almanack: The Wit and Wisdom of Charles T. Munger, Expanded Third Edition
  • Lia A. DiBello - Expertise in Business: Evolving with a Changing World (in The Oxford Handbook of Expertise) (via)
  • Joël Glenn Brenner - The Emperors of Chocolate: Inside the Secret World of Hershey and Mars
  • Elad Gil - High Growth Handbook/ audio
  • W. Edwards Demming - The New Economics for Industry, Government, Education/ audio
  • W. Edwards Demming - The New Economics for Industry, Government, Education/ the PDF or ebook
  • Henrik Karlsson - Escaping Flatland/ including the posts I SingleFile'd
  • the relevant-looking posts on benkuhn.net/posts
  • Commoncog Case Library Beta
  • Keith J. Cunningham - The Road Less Stupid: Advice from the Chairman of the Board/ audio
  • Keith J. Cunningham - The 4-Day MBA/ video
  • Cedric Chin's summary of 7 Powers
  • Akio Morita, Edwin M. Reingold, Mitsuko Shimomura - Made in Japan: Akio Morita and Sony
  • Nomad Investment Partnership Letters or redacted (via)
  • How to Lose Money in Derivatives: Examples From Hedge Funds and Bank Trading Departments
  • Brian Hayes - Infrastructure: A Guide to the Industrial Landscape
  • Accelerated Expertise (via)/ printed, "read Chapters 9-13 and skim everything else"
  • David J. Gerber - The Inventor's Dilemma (via Oxide and Friends)
  • Alex Komoroske - The Compendium / after I convert the Firebase export in code/websites/compendium-cards-data/db.json to a single HTML page
  • Rich Cohen - The Fish That Ate The Whale (via)
  • Bob Caspe - Entrepreneurial Action/ printed, skim for anything I don't know



Interactive fiction


unplanned notable things read


unplanned and abandoned

  • Ichiro Kishimi, Fumitake Koga - The Courage to Be Disliked/ audio
  • Matt Dinniman - Dungeon Crawler Carl/ audio
  • Charles Eisenstein - The More Beautiful World Our Hearts Know Is Possible/ audio
  • Geoff Smart - Who: The A Method for Hiring/ audio
  • Genki Kawamura - If Cats Disappeared from the World/ audio
  • Paul Stamets - Fantastic Fungi: How Mushrooms Can Heal, Shift Consciousness, and Save the Planet/ audio
@ivan
Copy link
Author

ivan commented Jul 30, 2024

Attachment as a driver of work will almost always produce good results. Unfortunately it may not produce useful results, or results that society wants. The Artist that aligns with society is rare.

https://www.eristicstest.com/the_artist/

@ivan
Copy link
Author

ivan commented Jul 31, 2024

  • If people have a good “user experience” when they interact with you, then they will want to interact with you more in the future.

https://dynomight.substack.com/p/advice

@ivan
Copy link
Author

ivan commented Aug 2, 2024

Ask HN: What is the best software to visualize a graph with a billion nodes?

Visualizing large graphs is a natural desire for people with lots of connected data. But after a fairly small size, there's almost no utility in visualizing graphs. It's much more useful to compute various measures on the graph, and then query the graph using some combination of node/edge values and these computed values. You might subset out the nodes and edges of particular interest if you really want to see them -- or don't visualize at all and just inspect the graph nodes and edges very locally with some kind of tabular data viewer.

It used to be thought that visualizing super large graphs would reveal some kind of macro-scale structural insight, but it turns out that the visual structure ends up becoming dominated by the graph layout algorithm and the need to squash often inherently high-dimensional structures into 2 or 3 dimensions. You end up basically seeing patterns in the artifacts of the algorithm instead of any real structure.

There's a similar, but unrelated desire to overlay sequenced transaction data (like transportation logs) on a geographical map as a kind of visualization, which also almost never reveals any interesting insights. The better technique is almost always a different abstraction like a sequence diagram with the lanes being aggregated locations.

There's a bunch of these kinds of pitfalls in visualization that people who work in the space inevitably end up grinding against for a while before realizing it's pointless or there's a better abstraction.

https://news.ycombinator.com/item?id=41132095

@ivan
Copy link
Author

ivan commented Aug 4, 2024

Intellectual property is absolutely a moat, but it's not the "Intellectual Property" as defined in law. It's the implicit, tacit knowledge & what exists in the collective mind of the team. But it's still property of a sort, the shareholders have a beneficial claim to it.

https://x.com/J_L_Colvin/status/1820119713654984913

@ivan
Copy link
Author

ivan commented Aug 4, 2024

Reinventing the wheel is a terrible metaphor in a lot of cases. I understand why it exists, but it seems to discourage trying something by yourself the first time. There is a lot of value in rediscovering something because you understand it a lot more, and in many contexts, that personal discovery is key.

https://x.com/inkolore_/status/1818739846774768064

@ivan
Copy link
Author

ivan commented Aug 4, 2024

@ivan
Copy link
Author

ivan commented Aug 5, 2024

Wow not a single mention of Whisper this entire comment first page! I think Whisper is really cool: the large model can pull speech out of even heavily distorted (wind noise, clipping, etc) audio. I have a story to illustrate why running Whisper on your own locally is not so easy! Much easier to sign up to the OpenAI API.

In my research I found that actually pre-processing the audio to reduce noise (using the IMO best-in-class FB research "denoiser") actually increases WER. This was surprising! From a human perspective, I assumed bringing up the "signal" would increase accuracy. But it seems that, from a machine perspective, there's actually "information" to be gleaned from the heavily distorted noise part of the signal. To me, this is amazing because it reveals a difference in how machines vs humans process. The implication is that there is actually speech signal that is inside the noise, as if voice has bounced off and interacted with the noise source (wind, fan, etc), and altered those sounds, left its impression, and that this information is then able to be utilized and contributes to the inference. Incredible!

With whisper: I started with the standard python models. They're kind of slow. I tried compiling python into a single binary using various tools. That didn't work. Then I found whisper.cpp--fantastic! A port of whisper to C++ that is so; much; faster. Mind blowing speed! Plus easily compilation. My use case was including transcription in a private, offline "transcribe anything" MacOS app. Whisper.cpp was the way to go.

Then I encountered another problem. What the "Whisperists" (experts in this nascent field, I guess) call it "hallucination". The model will "hallucinate". I found this hilarious! Another cross-over of human-machine conceptual models, our forever anthropomorphizing everything effortlessly. :)

Basically hallucination includes: feed Whisper a long period of silence, and the model is so desperate to find speech, it will infer (overfit? hallucinate?) speech out of the random background signal of silence / analog silence / background noise. Normally this presents as a loop of repeats of previous accurate transcribed phrase. Or, with smaller models, some "end-of-youtube video" common phrases like "Thank You!" or even "Thanks for Watching". I even got (from one particularly heavily distorted section, completely inaccurate) "Don't forget to like and subscribe!" Haha. But the larger models produce less hallucinations, and less generic "oh-so-that's-what-your-dataset-was!" hallucinations. But they do still hallucinate. Especially during silent sections.

At first, I tried using ffmpeg to chop the audio into small segments, ideally partitioned on silences. Unfortunately ffmpeg can only chop it into regular size segments, but it can output silence intervals, and you can chop around that (but not "online" / real time) as I was trying to achieve. Removing the silent segments (even the imperfect metric of "some %" of average output signal magnitude (sorry for my terminology, I'm not expert in DSP/audio)) drastically improved Whisper performance. Suddenly it went from hallucinating during silent segments, to perfect transcripts.

The other problem with silent segments is the model gets stuck. It gets "locked up" (spinning beach ball, blue screen of death style--I don't think it actually dies, but it spends a long, disproportionately long, time on segments with no speech. Like I said before, it's so cute that it's so desperate to find speech everywhere, it tries really hard, and works its little legs of during silence, but to no avail.

Anyway, moving on to the next problem: the imperfect metric of silence. This caused many issues. We were chopping out quieter speech. We were including loud background noise. Both these things caused issues: the first obvious, the second, the same as we faced before: Whisper (or Whisper.cpp) would hallucinate text into these noise segments.

At last, I discovered something truly great! VAD. Voice Activity Detection is another (normally) AI technique that allows segmenting audio around voice segments. I tried a couple Python implementations in standard speech toolkits, but none were that good. Then I found Silero VAD: an MIT licensed (for some model versions), AI VAD model. Wonderful!

Next problem was it was also in Python. And I needed it to be in C++. Luckily there was a C++ example, using ONNX runtime. (I had no idea any of these projects or tools existed mere weeks ago, and suddenly I'm knee deep!). There were a few errors, but I got rid of the bugs, and had a little command line tool from a minimal C++ build of ONNXruntime / Protobuf-Lite and the model. Last step was the ONNX model needed to be converted to ORT format. Luckily there's a handy Python script to do this inside the Python release of ONNXruntime. And, now, the VAD was super fast.

So i put all these pieces together: ffmpeg, VAD, whisper.cpp and made a MacOS app (with the correct signing and entitlements of course!) to transcribe English text from any input format: audio or video. Pretty cool, right?

Anyway, running Whisper on your own locally is not so easy! Much easier to sign up to the OpenAI API.

MacOS APP using Whisper (C++) and VAD0--conveniently called: WisprNote heh :) https://apps.apple.com/app/wisprnote/id1671480366

https://news.ycombinator.com/item?id=34992012

@ivan
Copy link
Author

ivan commented Aug 5, 2024

Apple Intelligence in 15.1 just flagged a phishing email as “Priority” and moved it to the top of my Inbox.

https://social.panic.com/@cabel/112905175504595751

@ivan
Copy link
Author

ivan commented Aug 5, 2024

“What are my options?”

Do you perhaps have a button to push… in case of an unexpected situation.

https://x.com/patio11/status/1817955327154831784

@ivan
Copy link
Author

ivan commented Aug 6, 2024

  • TOB-BREW-12, wherein a formula could opportunistically perform a privilege escalation through a user’s previously activated sudo token.

https://blog.trailofbits.com/2024/07/30/our-audit-of-homebrew/

@ivan
Copy link
Author

ivan commented Aug 6, 2024

In United States labor law, at-will employment is an employer's ability to dismiss an employee for any reason (that is, without having to establish "just cause" for termination), and without warning,[1] as long as the reason is not illegal (e.g. firing because of the employee's gender, sexual orientation, race, religion, or disability status). When an employee is acknowledged as being hired "at will", courts deny the employee any claim for loss resulting from the dismissal. The rule is justified by its proponents on the basis that an employee may be similarly entitled to leave their job without reason or warning.[2] The practice is seen as unjust by those who view the employment relationship as characterized by inequality of bargaining power.[3]

At-will employment gradually became the default rule under the common law of the employment contract in most U.S. states during the late 19th century, and was endorsed by the U.S. Supreme Court during the Lochner era, when members of the U.S. judiciary consciously sought to prevent government regulation of labor markets.[4] Over the 20th century, many states modified the rule by adding an increasing number of exceptions, or by changing the default expectations in the employment contract altogether. In workplaces with a trade union recognized for purposes of collective bargaining, and in many public sector jobs, the normal standard for dismissal is that the employer must have a "just cause". Otherwise, subject to statutory rights (particularly the discrimination prohibitions under the Civil Rights Act), most states adhere to the general principle that employer and employee may contract for the dismissal protection they choose.[5] At-will employment remains controversial, and remains a central topic of debate in the study of law and economics, especially with regard to the macroeconomic efficiency of allowing employers to summarily and arbitrarily terminate employees.

https://en.wikipedia.org/wiki/At-will_employment

@ivan
Copy link
Author

ivan commented Aug 6, 2024

97% of Americans' daily trips are done via automobile. Walking, biking, and bus riding tend to be associated with low socioeconomic status. There is heavy policing of low SES populations. In Kaplan, Louisiana it is explicitly illegal to walk at night. https://www.klfy.com/local/vermilion-parish/kaplan-starts-pe...

https://news.ycombinator.com/item?id=41105944

@ivan
Copy link
Author

ivan commented Aug 6, 2024

On July 1, 2021, Amazon did something it hadn’t done in many years: It added two new leadership principles to its list. “Strive to be Earth’s best employer” was one new principle. “Success and scale bring broad responsibility” was the other. 

When these were introduced, Amazon was facing intense media coverage over the swell of unionization efforts in its warehouse network, as well as antitrust and environmental scrutiny. And with Bezos scheduled to hand over the CEO reins a few days later, the introduction of the two new principles served as a convenient opportunity to demonstrate a company committed to progress under his successor.

Inside the company, though, the reaction among many employees was swift and brutal: “They are clearly a marketing ploy and they devalue the rest of the Leadership Principles,” one former senior manager of more than 10 years who left the company recently told Fortune. “Total bullshit,” opined a former longtime communications employee at Amazon who was still at the company at the time, and who summed up the new principles as “playing the reputational game versus guiding how you think.”

The problem, explained a former Amazon vice president, is that the pair of new leadership principles are so radically different from the other 14 that it makes it tougher for employees to grasp the essence of the overall principles and to act on them. “I’ve got to think that that’s somewhat challenging for the rank and file. The fact that there are so many LPs, and the fact that these last two are so abstract,” the former VP said.

https://fortune.com/2024/07/31/amazon-leadership-principles-questions-future-jeff-bezos-departure-andy-jassy/

@ivan
Copy link
Author

ivan commented Aug 7, 2024

@ivan
Copy link
Author

ivan commented Aug 7, 2024

@ivan
Copy link
Author

ivan commented Aug 8, 2024

Just for info: It is due to Sulfur in Cysteine.

Hi Gozenka, quick comment about the source of the smell. The smell in NAC doesn't actually come from the sulfur atom in the cysteine, pure cysteine does not smell. Rather, the smell is called by a plethora of small sulfur containing molecules that are created during the NAC production process and end up trapped in the crystals. During transit and use, the crystals break open and these small organic sulfur containing molecule volatilize and create the smell.

OP, if you want to make the smell go away you can take the powder, grind it up with a mortar and pestle to destroy all of the crystal structure, spread the powder on a cookie sheet, and just gently heat it in an oven or a warm room or in front of a fan. Once all of the sulfur containing impurities have volatilized off, the powder will not smell anymore. This is the easiest way to eliminate the smell, but there are others.

Hope that's useful info!

https://old.reddit.com/r/Supplements/comments/17friw8/nac_powder_really_smells_like_farts/

@ivan
Copy link
Author

ivan commented Aug 9, 2024

Accordingly, it was not spiritual illumination but everyday banality that he could not stand to lose. No line better distills the tender spirit of “The Book Against Death” than this: “Above all, when I am dead, what I will miss: the voices of people in a restaurant.” Canetti did not fret about the state of his soul; he worried himself sick about the fate of all the detritus that makes up a life:

What will become of all that has piled up within you, so much, so much, an enormous stock of memories and habits, deferred questions, frozen answers, thoughts, emotions, tender feelings, hardships, everything there, everything there, what will become of it all the moment life extinguishes within you? The disproportionate size of this stockpile — and all of it for nothing?

[...]

“He would like to die while writing,” reads one of his notes from 1986. “Before he’s entirely finished, he’d like to complete a sentence, exhale before the next sentence, and die exactly between the two.” And this is exactly what he did. His very last entry contains no acknowledgment that death is even possible. “It is time for me to sort matters out again within myself. Without writing I come undone. I sense how my life dissolves into dead, dull speculation when I no longer write down what is on my mind.”

https://www.washingtonpost.com/books/2024/08/02/book-against-death-elias-canetti-review/

@ivan
Copy link
Author

ivan commented Aug 9, 2024

The difference is that I just make claude-dev build very large pieces of functionality that I wouldn't be able to get done in cursor. I'm able to get it to build for example, a role based permission system on an enterprise crud app that already has a lot of functionality. I just tell it to do so and write the tests for it and then ask it 5 times to fix it to pass all the tests and it eventually does it without a me lifting a single finger.

https://old.reddit.com/r/ChatGPTCoding/comments/1embttz/is_claude_dev_finally_the_next_level_thing_we/?depth=99

I’ve been experimenting lots with claude-dev the last few days. It‘s hands down the most powerful agentic code generation tool I’ve used,by a significant margin.

Still lots of room for improvement but I‘m eager to see how much leverage the next generation of models will provide. Definitely already more productive as is than my usual workflow.

One aspect that might raise eyebrows is cost. A single days of intensive use on large codebases can easily cost about $80 to $140 for sonnet API calls.

https://old.reddit.com/r/ChatGPTCoding/comments/1en8dzu/state_of_gptcoding_aug_2024/lh53a6p/

@ivan
Copy link
Author

ivan commented Aug 9, 2024

Development velocity is the only variable to optimize. Poor stability reduces development velocity. Tech debt reduces development velocity. Pointless meetings reduce development velocity.

https://x.com/olson_dan/status/1820695663757455513

@ivan
Copy link
Author

ivan commented Aug 10, 2024

I interviewed a bunch of people who claimed enlightenment (or awakening/stream entry/similar shifts).

It seemed to me that there were different types of views on enlightenment among these people. Here's a bunch of ways they seemed to vary from each other:

The Bliss Spectrum: characterized by nirvana, ecstasy, constant happiness. They claimed they had reached a state where they were shivering with delight at all times.

The Mental Health Spectrum: of clearness, purpose-drivenness, fulfillment, being aware of what they wanted at all times, being motivated, healthy, whole. For them, ‘enlightenment’ was working all their personal shit out and having a hyper-functional life.

The Science Spectrum: Intense, awesome scientific epiphany. An awareness of their smallness, the universe’s largeness, deep realizations about being walking matter, of the intricate beauty of evolution.

The Superpower Spectrum: The possession of magic abilities, including being able to talk to beings from other dimensions, telepathy, and receiving sacred knowledge imparted to them from divine creatures.

The Concentration Tricks Spectrum: Intense concentration and mental abilities. They were able to alter their perception, slow time, change the space around them, see the refresh rate on monitors.

No-self spectrum: Altered senses of self; they included inanimate objects or other people in their sense of self, or had no sense of self at all. This includes ego death.

Understanding Spectrum: A general spiritual epiphany; people reported a sense of completion, of having ‘no more questions’, of finally comprehending what’s going on, of nonduality, of wholeness.

Wordless Spectrum: An altered or unusual relationship to words, typically distrustful or disconnected; playing with words, going meta with words, not using words, not-claiming, claiming contradiction and paradox.

Perception Spectrum: Seeing your experience as your own; experiencing yourself as inherent and integral to the experience; a realization that your brain is constructing things, often associated with dreamlikeness.

Morality Spectrum: An intense sense of an ethical direction; describes a ‘right’ or ‘moral’ direction, uses goodness as a consistent guiding description of their insights.

Love spectrum: A sense of love and compassion for all living beings

Tradition spectrum: How much they adhere to classic wisdom and beliefs; tends to believe in reincarnation and karma, places authority in the Buddha’s teachings

Disassociation spectrum: Stoicism; the ability to carefully control emotions, independence from environment, remaining unaffected by whims or pain or pleasure.

Peace: A sense of being deeply okay, at ease with the current moment, settled, present, like everything is all right.

https://x.com/Aella_Girl/status/1820932038238613797

@ivan
Copy link
Author

ivan commented Aug 10, 2024

These are models, supercomputer guesses as to what the weather might be. They are not realtime measurements of every square inch of sky.

The models are different, and have different goals and resolution.

  • GFS: Global, American Model, max 384 hours. Great for planning a few days out, but might not be as accurate in the 3-24 hour range, lower resolution (27km), updated 4x day. GFS27 and GFS+ variations.
  • ECMWF: Global, European Model, max 240 hours, 14km resolution, updated 2x day,
  • ICON: Global (ICON13), Germany Model, max 180 hours, 2x/day, 13km resolution
  • NAM: North America, US Model, max 60-84 hours, US-only, better resolution (12km), claimed to be BEST for North America, updated 2x day
  • HRRR: North America, NOAA Model, based on radar, good for short range, better resolution (3km), max 36 hours, updated 2x a day every 12 hours
  • RAP, WRF -- not in Windy, google them

https://old.reddit.com/r/paramotor/comments/y0znh5/dropping_the_windy_appneed_another_app_to_check/

@ivan
Copy link
Author

ivan commented Aug 10, 2024

These kinds of pesticide registration reviews are routinely delayed. Part of the problem, Glenna says, is that regulators have to rely on industries to release toxicology information because it’s considered proprietary information. “[It’s] intellectual property,” he says, “so university or publicly funded scientists simply don’t have the ability to do the research.”

When a pesticide is first registered with federal regulators, the vast majority of the information available about it is science conducted by the company who made it. “The presumption in the US is in favor of the safety of the chemical,” Burd says. Elsewhere, like the European Union, “chemicals are not presumed safe, they adopt a much more precautionary approach.”

In fact, according to federal law, the EPA can only refuse to register a pesticide if its risks are greater than the benefits it provides, as measured by crop yield or quality. As a result, close to a third of US pesticide use involves chemicals banned in China, Brazil, and the European Union.

There’s also a revolving door between the agency and the industry it regulates. Alexandra Dunn, the former assistant administrator for the Office of Chemical Safety and Pollution Prevention, for example, is now running CropLife America, the pesticide industry’s leading lobbying group. She’s only the latest; since 1974, all of the office’s directors went on to work for pesticide companies.

https://jacobin.com/2024/07/pesticide-cancer-lobbying-lawsuits/

@ivan
Copy link
Author

ivan commented Aug 11, 2024

When voting, there should be a separate box you can check marked "REBOOT" along with whatever other votes you cast. If over 60% of the voters check that box, all officeholders are disqualified from ever running again, and new elections are held the following year.

https://x.com/cmuratori/status/1822166430663774693

@ivan
Copy link
Author

ivan commented Aug 11, 2024

[MrBeast's] videos are mindless, crude, wasteful and contain zero educational or artistic value.

https://www.youtube.com/watch?v=NZu8-WnSqic

@ivan
Copy link
Author

ivan commented Aug 11, 2024

What's particularly alarming is the speed at which new vulnerabilities are exploited. In one case, attackers attempted to exploit a JetBrains TeamCity DevOps authentication bypass a mere 22 minutes after the proof-of-concept code was published. That speed is faster than most organizations can read the security advisory, let alone patch their systems.

https://www.zdnet.com/article/cloudflare-reports-almost-7-percent-of-internet-traffic-is-malicious/

@ivan
Copy link
Author

ivan commented Aug 11, 2024

The dilemma of the American conscience can hardly be better stated. Dostoyevsky was a great artist, and a radical one, but his early social radicalism reversed itself, leaving him a violent reactionary. Whereas the American James, who seems so mild, so naively gentlemanly—look how he says “us,” assuming all his readers are as decent as himself!—was, and remained, and remains, a genuinely radical thinker. Directly after the “lost soul” passage he goes on.

All the higher, more penetrating ideals are revolutionary. They present themselves far less in the guise of effects of past experience than in that of probable causes of future experience, factors to which the environment and the lessons it has so far taught us must learn to bend.

The application of those two sentences to this story, and to science fiction, and to all thinking about the future, is quite direct. Ideals as “the probable causes of future experience”—that is a subtle and an exhilarating remark!

Ursula K. Le Guin - The Ones Who Walk Away from Omelas

@ivan
Copy link
Author

ivan commented Aug 11, 2024

Users should be aware that the App Store collects exhaustive usage data and sends it to #Apple. This can’t be turned off. We made this video to show how tapping an app link gets recorded in details.

https://mjtsai.com/blog/2023/02/15/lawsuits-over-apple-analytics-switch/#lawsuits-over-apple-analytics-switch-update-2024-07-18

@ivan
Copy link
Author

ivan commented Aug 11, 2024

Erik Hoel, an American neuroscientist, posits that the industries AI are disrupting are not all that lucrative. He coined the phrase “supply paradox of AI” — the notion that the easier it is to train AI to do something, the less economically valuable that thing is.

“This is because AI performance scales based on its supply of data, that is, the quality and size of the training set itself,” said Hoel. “So when you are biased towards data sets that have an overwhelming supply, that, in turn, biases the AI to produce things that have little economic value.”

Hoel raises an interesting point. Generative AI’s current applications include writing, image and video creation, automated marketing, and processing information, according to the US Census Bureau’s Business Trends and Outlook Survey. Those are not particularly high value. Using specialist data, sophisticated models could do deeper scientific work, but that data can be in short supply or even restricted.

[...]

What I wanted to point out here though is actually something deeper. Which is that the longer AI takes to show up in any positive economic indicators, the more it becomes the case that AI has brought increasing existential risk in exchange for minimal upside.

https://www.theintrinsicperspective.com/p/all-the-existential-risk-none-of

@ivan
Copy link
Author

ivan commented Aug 11, 2024

Can other people free-ride on your skill? Can they look at your track record and say “everything this person picks goes down, so I should short it”? I mean! Tuttle Capital Management once launched some very rude exchange-traded funds to fade Jim Cramer’s and Cathie Wood’s stock picks, but I don’t think those had any real alpha; it is not actually the case that Cramer or Wood has this sort of anti-skill. In general, if your stock picks are known broadly, they’re probably not bad enough to make betting against you a reliable strategy.

On the other hand, your broker probably knows (1) what you’re buying and (2) how you’re doing. If you’re terrible, your broker might be tempted. Here is a funny CME Group disciplinary action against a futures broker called Wing Fung Futures Limited:

Pursuant to an offer of settlement in which Wing Fung Futures Limited (“Wing Fung”) neither admitted nor denied the rule violations or factual findings upon which the penalty is based, on July 16, 2024, a Panel of the Chicago Mercantile Exchange Business Conduct Committee (“Panel”) found that between July 22, 2020, and July 7, 2022, Wing Fung deployed an automated trading system (“ATS”) in the Australian Dollar, British Pound, Canadian Dollar, Euro FX, Japanese Yen, New Zealand Dollar, E-mini NASDAQ, Micro E-mini NASDAQ, and E-mini S&P 500 futures markets wherein the ATS’s strategy was to submit orders based upon trades executed by its clients who were regularly unprofitable. Specifically, upon a target client establishing a short/long position, the ATS generated a trading signal to enter an order in the same market for the same price and quantity as the client, though to establish the opposite short/long position. The Panel found that by employing this strategy, Wing Fung attempted to profit from its knowledge of its clients’ record of losses, which was not available to others in the marketplace. Additionally, the Panel found that Wing Fung employees used Wing Fung’s clients’ unique operator IDs to enter orders into Globex. The Panel further found that Wing Fung’s leadership fundamentally did not understand, or were otherwise completely unaware, of Exchange rules that prohibited the activity described above. Therefore, the Panel found that Wing Fung failed to diligently supervise its employees and agents.

https://www.bloomberg.com/opinion/articles/2024-07-22/bad-trades-are-also-valuable

@ivan
Copy link
Author

ivan commented Aug 11, 2024

Jump had a conundrum. The firm needed to test the mettle of its would-be staffers—whether they could parse the nuances in financial markets and translate them into algorithmic trading models. But it couldn’t give the temporary hires the keys to the kingdom, with its proprietary strategies and billions of dollars in capital.

Crypto offered a solution. The sector had its own tradable assets, exchanges, and quirks, but it was separated enough from Jump’s world of stocks and bonds that it wouldn’t pose a threat. “It was a bit of a toy market,” says one former employee, who spoke with Fortune on the condition of anonymity to discuss their previous firm. ...

https://www.bloomberg.com/opinion/articles/2024-08-05/the-good-trades-have-gone-bad

training/evaluation area

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment