Rephrasing a common quote - talk is cheap, that’s why I talk a lot.
Ah, in that dimension what I see seems similar to oil processing, again. They are generally all similar. Better datasets - better output. A natural curve of expenses and results.
A competitive open-source LLM makes sense ; but the real asset is data. So said LLM will be hosted (or provided with computing power) commercially to work on said processed data, usually. There are no anarchist free gas stations, and just like that it will be a building block of businesses.
It would be similar to an ethical LLM, but the question is not in ethics, it’s in having more structure. Sort of granularity. That could allow to scrape knowledge and reproduce it in some way better than just an LLM output. Such a thing could be both a model and an associative dictionary, a bit like automated Wikipedia.
I found it to be just Google made more convenient, which is good, but not there yet.
I know LLMs will get worse and shittier, which I think is a bummer, because they could be so damned useful.
Why would they? Humans keep producing new data. Old datasets will get less useful. They do all the time. And the old approach to training. But fundamentally they shouldn’t get worse.
Yes, so there was a time when I was dreaming day and night about something like those LLMs, but for archiving knowledge. That is, archiving existing statements with subjects and objects and relations, a bit more high-level and less generalized than LLMs. Syllogisms, semantic relationships, distances in application. Sort of what holocrons are in Star Wars.
The globe doesn’t have a left.
Yeah, so something’s profitability depends on relative extracted value from what it does. If your technology allows to, say, predict a person for 5-10 years forward in their decisions and reactions, the value extracted for that amount of calculations will be enormous. About what that person is worth.
And, note, this becomes more efficient with scale.
And gentle curve of scale is what this technology brings new to computing - data is like oil, hardware is like processing machinery, resulting predictions and extrapolations are fuel that makes those possessing it more powerful.
This is both more similar in application to early computers used for planning and air defense than much of computing history, and more accessible for comprehension by human mind without deep understanding of computing.
It’s a machine that answers questions based on data. It might sometimes answer them badly. But it answers questions better than a dice roll, and it does that for questions a human might not answer at all.
In other words, they are building a death star. Sleep well everyone.
Learning that paper borders are not borders and that second amendment you had was important, yes?
By the way, Americans in the interwebs like talking as if that second amendment were an outlier in the world.
Actually such concepts existed even in imperial China.
Even in USSR the founding myth postulated that it’s a revolutionary state and a revolution against reaction is a right and even a duty.
Except it works when everyone believes in it. When people start believing in paper borders and stop believing in power, they lose power.
I’m talking, of course, about weapons in wider sense. Scopolamine is a weapon. Even alcohol is a weapon. Knowledge is a weapon. Ability to process knowledge is a weapon. Predictive power is a weapon - what those big companies are investing into. Conditioning since kindergartens and schools is too a weapon.
The point is that everything in a society should be accessible and democratized and non-monopolized, because everything of value is a weapon, and things without value don’t survive evolution.
(Of course, evolution will work either way ; but that doesn’t support any strategy, because any choice at all will advance the bloodline, the population, the biome.)
A volunteer should be able to participate in everything, ideally.
There are some people who like hierarchies and “natural” violent relations, can’t decide if closer to Exupery’s Citadelle or to fascism. Sometimes very intelligent, but misguided. Those people would probably say I’m misguided and nature specializes.
Nature does everything. Nature specializes in small populations with limited connectivity, nature spreads and smoothes out differences in big populations, creates symbioses and different biomes. Nature does what’s best where it’s best and is perpetually being optimized for just that. Which is my point that they argue.
Because the system for going full Nazi has been built. For actual Nazis it was radio. Today it’s the Internet.
To be used for control it has to be regulated. It can be regulated so it will be.
Architecture defines outcome.
Yes! It’s reminiscent of Lem’s Ananke and Terminus for me, with illusions and inevitability of the former and feeling of soul in objects in the latter. Also there’s Eco’s Foucault’s Pendulum (which I still haven’t read in full, only in small pieces enjoying them quite a lot), addressing European occultism and fascism, which relays well a similar emotion that in fascism existed related to machinery, which was then new. Radio, automobiles.
Well, it kind of needs to be stupid at least. If it was smart it could talk back and then it loses its usefulness for the purpose of idolatry.
I think how we understand objects is important too. For the purpose of idolatry it’s sufficient to have only a small gap between functionality and understanding in the domain of will and choice.
Ancient fortunetellers looking at bird intestines were different from what their visitors expected only in that. Their visitors knew they want to learn what gods tell and not men, and that gods are not same as men, but more like the soul of the world around them. The only difference was will and choice, but these are infinitely small. One person can be predicted many years forward down to small things, if you learn enough about them. Whether they have will and choice is a question of metaphysics, in life it’s not resolvable. And it’s the same with whichever gods they believe in.
(And it had a functional role, a random decision is often better than one dictated by indirect application of interest.)
Well, that chord looks wrong, but I meant finally having a class of programs that works similarly to objects we encounter IRL and entities that human cultures are used to internalizing. And human cultures responding with acceptance.
Pygmalion is “Her (2013)” apparently.
Other than this I’m reminiscing on one of Lucian’s dialogues about a certain Aphrodite statue with extremely nice butt and one smitten visitor who was sneaking into the temple at night to pollinate that, resulting in precisely located mold spot.
Computers have finally caught up with humanity. This is good. I thought it’ll never happen that they are finally a part of human magical thinking. This is as terrifying as it’s inspiring.
So, from Russia - let’s see how people in your country of “dignity and culture to resist oppression and deep institutions” will pass through that future. From someone “not protesting enough” and “feeding the regime with my taxes” and “feeling fine when Ukrainians suffer” (technically true, except might not be the only thing true).
Yes, so my well-trained ears prefer noisy sound, something like 48kb mp3s I downloaded from the web in my childhood (born 1996). Because that’s less likely to cause migraine through them than a good record with some annoying sounds in it, preserved by a more precise lossy encoding. And things you want to hear are kept well enough even by 48kbit mp3s.
And this surprisingly keeps with analog things, like headphones and speakers. I prefer something cheap and noisy that makes sounds softer to something quality and with crisp sound, but somehow too crisp.
And I do have good ears, I can hear a lot of things, a cat walking on a neighboring plot in countryside during wind, things like that. Hence the migraines.
That’s because a social app that quickly solves the need of making a connection and then perpetually the need of maintaining it was called ICQ. Or AIM. Or other such. They were focused on the part after that hope.
The reason that’s no longer the normal model is simple - weak people are easier to exploit. The “after hope” model doesn’t keep people weak.
Even with XMPP - the classic instant messenger model of adding someone to friends, remember it? You send one invite message, and after it the other side won’t see anything you want to send until it accepts you into contacts. It might never do that. Or it might add you, see you’re sending unsolicited dick pics, remove you.
With dating apps all you need is a search by tags and tags corresponding with truth, and of course ability to choose who can contact you. The former is not hard. The latter is hard when people are interested in putting false tags, but not when the tag social metric, so to say, is commutative. The model where conversations are started by mutual “like” is good, I think. And the anonymized way (like with Pure, have tried using it when decided to become more social, got some insights but no dates, or more specifically one failed date) is good, when those who liked you are shown as anonymous invitations to accept or deny, but also when mutual “like” means accepting that invitation. I think one’s visibility and one’s point of view are something that should both be customizable with logical conditions. One should be able to set they want to only be seen by people without “no less than 20 inches” wish to not be frustrated when those people ghost them, or that they don’t want to see people without photos on their page, or that they only want to see people and be seen by people who like hiking or who like animals, but not both at the same time, or any other set of logical rules, everyone is different. Perhaps a limit on searches is good, though.
And then there is crime. Or mental illnesses. Or bad hygiene. Or conflict. That is, there are situations where outside observers should be able to evaluate who of the two sides is telling the truth about the other side being an abuser or whatever. I suppose some kind of escrow for contacts can be devised. This should be a social thing, a moderator can’t be trusted with correspondence and also with judgement. So - escrow by people trusted by both sides, something like that. To have a rating, it should be possible to tell who’s really spilling tea and who’s doing libel.
And if you were reading attentively, you might have noticed this doesn’t just apply to dating, this applies to everything about establishing contact over social media. Because that’s absolutely correct, dating doesn’t differ in anything from any other social connectivity. In other social events you too want to quickly find and communicate for long with someone. Romance being involved doesn’t change much or anything.
The reason these two purposes have been separated by businesses is pretty transparent - trying to apply general social media to dating shows that they don’t work, and trying to apply dating social media to normal long-term communication shows that they too don’t work. The issue is that what’s invisible still exists. That separation is just hiding what doesn’t work, but it still doesn’t work. A functional social media would function for both dating and daily buddy talk. Like ICQ did.
because if they serve that official goal, they’re essentially nixing their own customer base.
Not if it’s a promiscuous sex dating app. Then it’s more of a social network, thought, just around polyfuck graph instead of kitty photos graph.
As the poster above, I should clarify that the reason I mentioned Telegram is that there a channel (like a blog) has a representation as a group chat where channel posts, comments to them and simple group chat messages appear.
And about communities and issues - the problem with comments is that they are local to post. Separation by posts first, then separation by threads, separation by score ranks, separation by depth. That may seem like a nice idea to not see everything. That’s the very problem.
OK, so the data model wouldn’t have to be changed to make it in good sense like Telegram.
What you need is ability to have a linear representation, where every message is additionally marked as a comment to some post or as a reply or as a post itself, and might have scores.
Like old forums usually were, tree representation wasn’t very popular.
Comments: that thing where you upvote what you like and downvote what you dislike, and if you feel strongly, say the same thing 10 other people said in different words. Very important. Very necessary.
Honestly a sidebar with a chat like IRC, but every message being tagged by which specific article its author is viewing right now, would help the social part more.
I mean, Telegram’s convenience is a weapon. If you want a good UI for article feeds with comments, it should just copy Telegram. Except for the Russian intelligence services part.
I’m talking about this not being great depression levels.
It’s as if you weren’t reading what I’m saying.
That aside, roasted sunflower seeds are nice, it’s the stereotype of young pigs littering their shells in, eh, every public place where they are consumed, that’s a negative one.
Also roasted sunflower seeds are a product of Ukraine and south of Russia mostly, so the stereotype is as suited for Ukrainians as for Russians, perhaps more.
A minimum wage job, 1 minimum wage job, paid for a family in 1950. Bought a house, a cheap car, doctors, dentists, optometrists, other professional services people can’t do themselves, even able to go out to eat for a burger.
Yes, in 1950, damn right. Now do 1930.
by the 1980’s it was undeniably true that wages no longer paid for what they did
Still better than 1930.
but there is no future going further up, come straight back and join us in the sunshine brother!
I dunno, there might not be any sunshine stored for me, but it’s still not 1930.
Well, if we continue my analogy, government-run oil processing plants and gasoline subsidies have not historically worked well.
It’s a device of investing hard power into computing.
That cropland will repurpose itself by market laws if the change is so dramatic, I think it is. I don’t like the AI hype, but the major change of converting hard power into data and data into answers to questions is potent enough. It’s not just the difference in energy volumes between ethanol and solar power, it’s also that liquid fuel is easier to store. It’s not an equal comparison you’re making. But if the energy demand is skewed enough on the side of grid-connected datacenters, then economically solar power might become more attractive.
I think oligopoly on data is the main threat to this. Datacenters and hosters providing power to run whatever you want with whatever data you want are not the bottleneck for competition and good evolution.
Various data harvesting farms in which users roam are.
It’s funny, I’m optimistic lately and feel like this family of technologies is slowly killing the oligopolies of previous generation. Well, not themselves, but the mechanisms that brought them into existence. Of course they too have moved on past those, but it’s sort of an improvement.