@glyph Did you quote post something?

[miss_rodent@girlcock.club]

@xgranade @glyph Yeah, I think there is a lot more potential to QC than there is to LLMs, and likely to AI, overall, though machine learning and stuff, more generally, have some very actually-useful applications in various fields. Trying to make a chatbot into a new work-eliminating machine god seems like a pretty obviously hopeless endeavor though.
I don't think the useful parts of either are going to earn the trillions that could salvage this sort of datacenter money-burning craze though.

[xgranade@wandering.shop]

@miss_rodent @glyph I spent about 10 years on classical ML tools for building QCs and another few on using QCs for doing ML, with a bonus serving of "using small QCs and novel ML algorithms to build larger QCs." Needless to say, I agree about there being some useful stuff under the moniker of machine learning.

[glyph@mastodon.social]

@xgranade @miss_rodent love the genre of mastodon post which is "as [literally the most qualified person in the world on this specific topic], 'yes'"

[astraluma@tacobelllabs.net]

@glyph I wonder if the industrial revolution was like this--a few years of wild continuous change and then settling down into a stable form.

[glyph@mastodon.social]

@astraluma I don't think any of it narrativizes quite that precisely and neatly, and it depends exactly where you were and what you were doing at the time, but, in general, I think, "yes"

[miss_rodent@girlcock.club]

@glyph @xgranade On the plus side, the 'extremely qualified yes' genre of post is at least a good sanity check on my side of it, I'm not nearly as qualified and have shifted focus more into philosophy and the humanities since dropping out of my math/physics major in uni .

[triplefox@pounced-on.me]

@glyph thinking about how this has played out in my lifetime, I can think of three key ingredients that probably repeat throughout history and produce "Big Thing" effects: presumption, speculation, and coercion.

You need the presumption to start thinking about a technology in terms of large scale adoption - a "computer on every desktop" kind of statement has to be collectively believed enough for an investment market to exist. Everything SV pushes is an attempt to rewrite history and develop new idols; startups resemble cults because their employees have to believe in something radical that is often unpopular when related in its full consequence - e.g. spreading computers means you've invented computer illiteracy as a social penalty. "But computers are cool so whatever," we say looking from the inside.

The investment has to be speculative because the new presumption has to counter a previous one, like "there is a world market for maybe five computers". One challenges the other in the marketplace. If it's an executive change then there's no power struggle to make the thing "big".

And the market has to be supportive of gangster dynamics in some fashion, where the speculative entities coerce each other and other industries into consolidated forms - in kinetic confrontation on a battlefield, in courtrooms and contracts, through marketing and manipulation of common vocabulary and idioms. If the change occurs without a coercive push then it doesn't manifest big opportunities to grab wealth.

Right now, the first is the one that is failing the most frequently. There is investment and coercion but it's not towards believable ends, which redirects the resulting activity towards "loot and plunder." Rises in authoritarian thinking signal that the prevailing dynamic of the powermonger and how they engage in the economy has shifted from "own relatively less of more" towards "own relatively more of less". So they kick the legs out from underneath a lot of older presumptions to try to get something going for themselves and we get a nihilistic sentiment from the top.

In this period I believe artists and philosophers may hold a position of cultural strength because training towards "different ways of seeing and thinking" also develops novel belief systems. When the culture is being stuffed with a battle for Big Things, nobody has time for it - they are trying to stay in line and take the easy opportunities, socially speaking, so they praise the incumbent winners or the managed competition of celebrities, sports, and political parties, and we get a monocultural dynamic. But right now, monoculture is dead - or maybe it's just sleeping.

[gimulnautti@mastodon.green]

@glyph The enrichment of the 1% or 0.1% is effectively inflation. What’s actually going on is that managent of people living in poverty is becoming cheaper and cheaper..

[carlton@chaos.social]

@glyph “Front-loading a massive amount of conceptual complexity to accommodate the problems of hyper-scalers is a common mistake.” 🥳

[leah@blahaj.social]

@glyph @bitprophet yet we are only 24% into the century

[neal@social.gompa.me]

@glyph It's a great blog post, but I think we often recognize "big" things in retrospect rather than as they are coming. Nobody really considered the Internet "big" when it was first being put together. Nobody considered Google, Facebook, Twitter, etc. to be "big" either.

The behavior of VCs is based on this evidence. And I don't think they're wrong for it either.

Maybe this is it for "big things", maybe not. Who knows? Just enjoy the ride for the next century, I guess. 😄

[neal@social.gompa.me]

@glyph To be clear, your general advice about building small-scale things is still good. It makes everything a lot easier. Simplicity ironically often leads to clean and scalable software.

(Sometimes the problem domain makes that hard, though. 😦)

[lynnntropy@hachyderm.io]

@glyph It's obviously a very different kind of tool, but Homebrew solves this problem, no?

[glyph@mastodon.social]

@lynnntropy Nowhere near as comprehensively. Relatively little software is available as casks.

[mi@awawa.cat]

@glyph i keep thinking an example of things built for a too-large, non-personal scale is current state of Unix-likes and main distros thereof. It seems to me we now once again lack readymade installable setups suitable for hardened personal/small-business VM deployment, and we keep requiring lots of expert knowledge from anyone choosing to nowadays exist independently in the hostile web. Also replied on it elsewhere: awawa.cat/notice/B1qyfkwUVjGenI9fNI

[bitprophet@social.coop]

@leah @glyph the last 5 years easily count for at least 10 tho!!

[snoopj@hachyderm.io]

@glyph the main things I would have remarked on in this vein would be that footnotes 2+3 seem out of order to me, and ZIRP isn't defined (or a definition linked to)

Otherwise? Another banger. Especially like the call to action to be prepared for the free-money party in "AI" to come to an end, it's wild seeing how few developers going in for the technology are acknowledging this inevitability.

[glyph@mastodon.social]

@SnoopJ thanks for the feedback, I did in fact fail to run my footnote fixing script this time. Not sure if I should update it given that it would swap the link targets now … but I can add an Investopedia for ZIRP or something:)

[snoopj@hachyderm.io]

@glyph yea, I wouldn't go to any trouble to adjust it, but I did notice it :)

[jmeowmeow@hachyderm.io]

@glyph Isn't "Story 1" (small team, speculative product) the classic case for portfolio funding across multiple bets, perhaps in some product area where your investors have expertise?

When a bet fails or doesn't zoom upward, you know something more about the management and staff and whether you might want them on a new project, or perhaps you have a story for merging overlapping proposals to better cover a segment. (my previous employer was two start-ups merged by private capital and sold to a larger company as a unit).

Story 2 does seem like "too much capital, not enough investment staff and expertise" and it leans toward cornering territory rather than creating new value.