raze2012's recent activity
-
Comment on How can I completely and permanently remove the ability to access the internet from a Debian derivative? in ~tech
-
Comment on Google blocks some California news as fight over online journalism bill escalates in ~tech
raze2012 Link ParentFrom my understanding here, the actual payment details are all very vauge (which I imagine is by design. How often do exact dollar figures age in history?). Google et. al needs to allocate some %...is the fee based on search result hits or search result/news summaries where the source is the news outlet? Both?
From my understanding here, the actual payment details are all very vauge (which I imagine is by design. How often do exact dollar figures age in history?). Google et. al needs to allocate some % of their ad revenue from search to paying news outlets, which is divied up based on some other sort of process to determine how much each site gets. Those divies per site can be from objective metrics like traffic/how much Google uses summaries from which site, or based on completely subjective factors like site reputation, quality, projected reach, etc.
I imagine there will be a lot of back and forth over those numbers and divy methods if Google does manage to relent in their current blocks.
-
Comment on Google blocks some California news as fight over online journalism bill escalates in ~tech
raze2012 Link ParentHere's the raw bill for the lawyer types. for me and other laymen, this summary seems to suffice: Summary of summay. So yeah, basically part of the ad revenue goes to journalism websites on some...Here's the raw bill for the lawyer types.
for me and other laymen, this summary seems to suffice:
Summary of summay.
This bill, the California Journalism Preservation Act, would require, within 10 days of the close of each month, quarter, a covered platform, as defined, to remit a journalism usage fee payment to each eligible digital journalism provider, as defined, that submits to the covered platform a certain notice. Under the bill, the journalism usage fee payment would be equal to a percentage, as determined by a certain arbitration process, of the covered platform’s advertising revenue generated during that month quarter multiplied by the eligible digital journalism provider’s allocation share, as defined, for that month. quarter....
This bill would prohibit a covered platform from retaliating against an eligible digital journalism provider for asserting its rights under the act by refusing to index content or changing the ranking, identification, modification, branding, or placement of the content of the eligible digital journalism provider on the covered platform.
So yeah, basically part of the ad revenue goes to journalism websites on some case by case basis, and some general protections against retaliation by search engines.
And I suppose in the name of fairness, I'll present Google's official argument and interpretation of the bill
Counterargument summary
>By helping people find news stories, we help publishers of all sizes grow their audiences at no cost to them. CJPA would up-end that model. It would favor media conglomerates and hedge funds—who’ve been lobbying for this bill—and could use funds from CJPA to continue to buy up local California newspapers, strip them of journalists, and create more ghost papers that operate with a skeleton crew to produce only low-cost, and often low-quality, content. CJPA would also put small publishers at a disadvantage and limit consumers’ access to a diverse local media ecosystem..... > >For more than two decades, we’ve provided substantial support to help news publishers navigate the changing digital landscape and innovate. We’ve rolled out Google News Showcase, which operates in 26 countries, including the U.S., and has more than 2,500 participating publications. Through the Google News Initiative we’ve partnered with more than 7,000 news publishers around the world, including 200 news organizations and 6,000 journalists in California alone.Their main point against this (outside of he profit motive) has to do with this mostly benefitting the larger publishers, and how this may even encourage them to consume smaller ones as they essentially get funded by getting views (which they are already good at). There definitely should be some mechanism to keep this from further skewing journalism.
-
Comment on Google blocks some California news as fight over online journalism bill escalates in ~tech
raze2012 Link ParentThe concept is that Google these days scrapes and summarizes the news pieces, which proceeds to disincentivize actually clicking on the link. So it gains value from extracting value out of its...The concept is that Google these days scrapes and summarizes the news pieces, which proceeds to disincentivize actually clicking on the link. So it gains value from extracting value out of its sources, a parasitic relationship instead of a symbiotic one.
This will only get worse as AI proceeds, so this may have more ground today than a few years back when Canada attempted this notion.
-
Comment on Texas is replacing thousands of human exam graders with AI in ~tech
raze2012 Link ParentTraditionally, someone well off enough and influential enough tends to need to light the spark, though. Not meaning to accuse you, just noting that often the people struggling also can't vouch on...and I think such a movement needs to be started by people worse off than me. It’s kind of hollow if I start complaining on other people’s behalf.
Traditionally, someone well off enough and influential enough tends to need to light the spark, though. Not meaning to accuse you, just noting that often the people struggling also can't vouch on their own behalf. Be it due to lack of awareness of what they can do, lack of energy, or in general lacking the voice of reach.
On the contrary, you definitely see people conned by other people much better off. So if you look at it less as "I need to be part of the community struggling" and more of "I need to be as well off as the conman and keep them in check", you may understand why this happens historically.
-
Comment on Texas is replacing thousands of human exam graders with AI in ~tech
raze2012 Link ParentThat's another part of the issue. Everyone learns and approaches learning differently, but public education strives to make a general program. And then public education approaches change from...So with the same intensity that I will make political decisions I will protect my approach to learning. No bullshit is accepted.
That's another part of the issue. Everyone learns and approaches learning differently, but public education strives to make a general program. And then public education approaches change from primary school to secondary, which can cause further clashes. The motivated ones will adapt no matter the situation, but other will simply hit their heads against the wall and be punished for "not being smart". I won't deny a bit of this is due to the student themself, but there's a massive burden for the system to carry. And a moderate burden on the home situation of each child.
The answer seems obvious, but the US's "No Child Left Behind" initiative proved to be a horrible decision. I don't really have any simple solution here, honestly. Especially for younger students that may lack the maturity to choose between something like a "exam track" or a "study track" in terms of how grades are weighted.
-
Comment on Texas is replacing thousands of human exam graders with AI in ~tech
raze2012 Link ParentThat's part of the reason there's been no mass revolt. Enough people are doing just okay enough to be satisfied in conformity rather than fighting back to try and get some leverage in the job...I know it’s bad out there and I’m really lucky to be where I am.
That's part of the reason there's been no mass revolt. Enough people are doing just okay enough to be satisfied in conformity rather than fighting back to try and get some leverage in the job market. It's part of why economic discussions can be so caustic.
-
Comment on Texas is replacing thousands of human exam graders with AI in ~tech
raze2012 Link ParentAs a devil's advocate, the student doesn't necessarily always know they understand it. I definitely had way too many times I thought I understood the subject, then I do a problem slightly off...As a devil's advocate, the student doesn't necessarily always know they understand it. I definitely had way too many times I thought I understood the subject, then I do a problem slightly off kilter and I couldn't connect the concepts together on a test.
And yes, I think that college gap is part of why a lot of high school teaches to skew homework over exams. You may need to repeat some content anyway, so the core point isn't about mastering a topic, it's mastering how to learn. Once you hit college, they barely care about homework.
-
Comment on "PS5 has no games" in ~games
raze2012 Link ParentThat's the rub, "if you own a PC (with your desired specs)". Most don't. And these days the replacement cost won't exactly outpace a console. You replace a GPU in 7 years around the PS6 release...because a PC of comparable specs is a substitute
That's the rub, "if you own a PC (with your desired specs)". Most don't. And these days the replacement cost won't exactly outpace a console. You replace a GPU in 7 years around the PS6 release and it might be cheaper, or might be the cost of the console.
IMO with the current PC market it comes down to preference more than savings. I use a gaming laptop for development and occasional PC gaming, but I prefer my Switch and PS5 when I could for gaming.
This is true but irrelevant - if you only target one console, then you're sacrificing at least 2/3 of the market (and profits).
Okay, that's the usual ebb and flow of development. Sometimes if you chase two rabbits you get none. You can have a great game ruined or compromised because you couldn't handle all the console bugs before ship, or because a low powered target like the Switch limited graphical techniques you could use.
Despite the outcry, you can indeed work on another port after the first one proves itself. People won't refuse to buy your game because their platform of choice wasn't first priority. I'm not a fan of port begging, at the end of the day, people wanting to have your game on their platform is a good sign, not a bad PR sign to bet it all in one go next time.
Also, having a single platform for devs to target isn't exclusive to consoles - the Steam Deck is "not a console" (you can install whatever on it) and yet is a single platform for devs to target.
Sure, but that's a spec target, not a console target. The whole allure of Steam Deck is that you can run a windows build on it and test accordingly. So it's less "test steam deck" and more "test steam deck specs", ideally only hitting a few quirks to quickly adjust.
If Pc Devs had to target and test every little spec, they'd go insane (and waste a lot of time. a 3080 and a 4080 aren't going to have specific differences outside of slightly faster performance).
but that argument goes both ways - consoles selling very-stable-experience boxes at a loss absolutely destroys the market incentive for gaming PC sellers to produce a very-stable-experience box, because they can't compete on upfront price. We don't actually know whether the console incentives are necessary here.
Yeah, sounds like good business for console developers. Undercut the competition and promise support and stability. A tactic as old as time. You were mentioning that you can replace a PS5 with a similar specc'd PC, but as of now the pricing optics + stability are good arguments for getting a console.
I also feel like you're missing the forest for the trees, here - while I'm sure there are arguments for consoles here and there, when they exist they're weak and the important question is whether they're strong enough to justify the drawbacks of the anti-competition.
With all due respect, this perspective is extremely myopic. I'm a dev so clearly I can modify and tinker with my computer until the cows come home. And I've done it dozens of times.
But some 95% of the population cannot, does not know how to, nor wants to do that. As long as convinience is preferred, there's a market to tap into. That's how Mac and especially iPhones got their traction, that's how popular social media got their traction. That convinience in and of itself is the answer and justification. You can call it weak, but you're ultimately making a judgement call on people who simply have different priories in their lives. To each their own.
I don't really see the anti-competiton angle either. If Pc really is "2/3rds of the market" then there's not much angle to stand on just because it doesn't literally have 100% of the games you want to personally play. A console manufacturer choosing to release a game on the console they made to justify its console doesn't seem anti-competitive. A business choosing to take a deal to release on a console first isn't anti-competitive, especially when some studios historically stagger their ports anyway (several Japanese studios employ this).
if you really want a stable development platform on PC then you could just target an emulator.
I imagine there are too many legal and logistical issues with this (let alone technical quirks) to ever risk this. Devs need to sign contracts and work with dev kits and throwing out a rom that works on a modern system emulator seems like easy lawsuit fuel (probably. I haven't looked at that contract in a while).
-
Comment on From its start, Gmail conditioned us to trade privacy for free services in ~tech
raze2012 Link ParentI wouldn't say gmail got worse, but they definitely removed features and options that some users enjoyed. for mobile apps, many swore by Inbox, but Google shut that down while it barely came out...The competition has gotten better since then, but I don't think Gmail has gotten worse?
I wouldn't say gmail got worse, but they definitely removed features and options that some users enjoyed. for mobile apps, many swore by Inbox, but Google shut that down while it barely came out of beta.
But the bigger issue is that consumer email is more an obligation than a "valuable service". I need an email for pretty much any website I actually want to use. I basically get an email if I want to use any Google service whatsoever, from Google Drive, to youtube, to Fitbit or any of the dozen services they acquired over the years. And emails are semi-required for contact if you are applying to any job (I wouldn't be surprised these days if college applications ask for an address).
None of these are thing I want to have to do. like physical mail I check it, clear out 90% spam, check out the 1 auto-generated rejection letter, and move on. Not necessarily something I "value" per se,
-
Comment on From its start, Gmail conditioned us to trade privacy for free services in ~tech
raze2012 Link ParentFree as in beer =/= Free as in freedom. But I think the bigger point was that there was no alternative premium email service to use if you cared about privacy. There definitely isn't a semi-known...Free as in beer =/= Free as in freedom. But I think the bigger point was that there was no alternative premium email service to use if you cared about privacy. There definitely isn't a semi-known premium email service to this day, Plenty of more niche services, but nothing really caught on.
-
Comment on Hey, monthly mystery commenters, what's up with the hit-and-runs? in ~tildes
raze2012 Link ParentI think those users are simply infrequent viewers. They may pop in, make a few comments, and then not think about Tildes for days. By the time they see a reply, the window is long gone to reply....That's the thing, they often are not. That is what made me scratch my head. They are often comments like this one. Specific replies aimed and relatively short.
I think those users are
-
simply infrequent viewers. They may pop in, make a few comments, and then not think about Tildes for days. By the time they see a reply, the window is long gone to reply.
-
they have a pattern or internal ruleset to never/seldom respond to replies. As said above, could be old cruft from less charitable communities who seem to be rife to get into a fight rather than have a conversation. Could be a habit from super large websites where you will almost never get a reply (try making a comment on a youtube video older than a few hours on a channel with >100k subs, or especially in a stream with hundreds of viewers. You grow to never expect a back and forth)
-
-
Comment on Noam Chomsky: The false promise of ChatGPT (gifted link) in ~tech
raze2012 Link ParentBe it a database, an ancient tome, a brain, or a markov chain, we need to have some mechanism to store events, and a mechanism to recall the events. the structure and medium may be different, but...For one, again you're not storing data.
Be it a database, an ancient tome, a brain, or a markov chain, we need to have some mechanism to store events, and a mechanism to recall the events. the structure and medium may be different, but the effect is the same. If it's not actively scraping information in real time and discarding it, it is storing some data, somehow.
I'm not really suggesting that it literally stores or scrapes data because I don't have enough knowledge of the architecture to make such a call. Plus, the underlying details of such storage isn't really of interest outside of ongoing legal battles (and even then, it's a footprint among the real issues of LLMs as of now).
I only mentioned video games to say that the size of the weights is not a computing bottleneck for LLMs.
If it's not a lookup table (since you emphasize how this is in fact not a table), it's bound to take a non-trivial amount of time to compute the result to respond with. Sure, 80GB is trvial in an age of big data that crawls through hundreds of terabytes a day in any given company. But we still have to operate within the realm of physics. a CPU only does so many operations, so if you can't brute force it, you either need to filter down and cache the data, or get more novel techniques to traverse the data with.
-
Comment on Noam Chomsky: The false promise of ChatGPT (gifted link) in ~tech
raze2012 Link ParentIt's not 1:1 with a traditional database of course, but the fundamentals of it comes down to "store data, search for relevant data," and then on top of that attempt to interpret the relevant data....I wouldn't say it's checking anything, nor is any cleverer
It's not 1:1 with a traditional database of course, but the fundamentals of it comes down to "store data, search for relevant data," and then on top of that attempt to interpret the relevant data. The data doesn't comes from navigating the natural world, so something needs to be provided, and then corrected when it inevitably gets misinformation. Or simply receive curated information to begin with to minimize corrections. That's part of why LLMs are in a legal battlefield as of now.
The size of its weights is only 80gb, which is about what a modern video game is.
modern video games are mostly textures and sounds taking up that space. And games aren't usually doing anything particulaly novel with the assets, just caching them where needed and replaying data that is cached. Add in some minor simulation for physical contact and you got a game (on a bytecode level, at least)
I imagine an LLM is storing a lot more textual data as well as compressed assets in certain products. It's doing a lot more with that data than streaming it so you need a mix of better algorithms to reduce traversal, and better hardware to speed up the traversal. That's why a game can be optimized aggressively to run at 60fps, but you can easilly take seconds to filter through a mere thousands of nodes of data naively. 80GB of such data means billions of such nodes to process.
-
Comment on What are some of your favorite PlayStation 1 games? Any odd or unique ones worth playing? in ~games
raze2012 LinkI'd definitely 2nd Kouldelka if you want that peak "PS1 era" style of game (for better or worse). I'll just use this moment to remind that this game leads into the Shadow Hearts series if you want...- I'd definitely 2nd Kouldelka if you want that peak "PS1 era" style of game (for better or worse). I'll just use this moment to remind that this game leads into the Shadow Hearts series if you want to see more of it later on.
- Parappa is relatively well known so not quite a "unique" experience for a PS1 gamer, but it's definitely one of those kinds of games you can't imagine a large studio creating today.
- There are loads of licensed games in this era, and most (as you'd expect) are awful. But you can probably find a few gems if you search around. I recall a Toy Story game and a Spiderman game being really good.
- A personal favorite of mine that was lost in the wave of RPGs is Threads of Fate. another one of Squaresoft's many JRPGs, this was a 3D-ish (all 3d, but level progression feels like a side scroller half the time) action-RPG that featuring 2 POV's from 2 protagonists, with two separate campaigns. A kind hearted male adventurer who can transform into monsters he encounters, and a greedy female mage who has a mix of close and ranged magic combat prowess. I can understand why Final Fantasy 7-9 eclipsed it, but it holds a place in my heart for its more light-hearted charm. If you liked Megaman Legends, this is probably right up your alley in terms of vibes.
any weirdo shit i should play?
oh, if you want peak weird, you may as well experience the glory of Pepsiman with your own two eyes. Truly a product of its time.
-
Comment on Noam Chomsky: The false promise of ChatGPT (gifted link) in ~tech
raze2012 Link Parentas a counterargument, artists have been making social commnentary about the impact of technology decades before any of the ambitions could be realized. Sci-fi authors of the mid 1900's would...as a counterargument, artists have been making social commnentary about the impact of technology decades before any of the ambitions could be realized. Sci-fi authors of the mid 1900's would barely live to see the first personal computer, but were writing about the (usually cynical) implications of advanced tech that we ourselves in 2024 may barely live to see.
None of us know the future, but we can indeed influence the future through such conversation. I think "listening to non-experts" is worth at least that much compared to shutting down the conversation to a specific lens.
-
Comment on Noam Chomsky: The false promise of ChatGPT (gifted link) in ~tech
raze2012 Link ParentIt is still "checking stuff", it's just more cleverly stored than "5 billion pages of tumblr pages". It ultimately still does come down to databases, though. that's why a lot of the bigger LLMs...ChatGPT isn't checking anything. 5 billion pages of Tumblr pages imprinted the statistical connection between their words in ChatGPT's weights. It does not have a copy to reference to.
It is still "checking stuff", it's just more cleverly stored than "5 billion pages of tumblr pages". It ultimately still does come down to databases, though. that's why a lot of the bigger LLMs can't simply be installed on a consumer computer (outside of companies not wanting clients to reverse engineer their golden goose). It still needs massive stores of servers to properly resolve its "inferences" in an accurate (and slightly untimely, but ultimately responsive enough) matter.
. If I forget something, I can remember and dredge things back into short term memory. If a word connection is too far gone in ChatGPT's weights, it's outta there.
I'll leave the somewhat obvious disclaimer here that our recollection isn't perfect. In some ways, ChatGPT hallucinating may be the most human aspect of its operation. Sadly, market forces and culture mean that these companies would rather an LLM try to act like it knows something more often than not as opposed to admitting its folly and admitting it cannot follow.
-
Comment on Noam Chomsky: The false promise of ChatGPT (gifted link) in ~tech
raze2012 Link ParentWe are still far from uncovering the full power of the brain, but I can see this point being true in some aspects. We process a lot of information, but we don't have all that information stored in...Now, is that really true? It seems to me that humans are processing a pretty vast amount of information. At a gigabyte of information per hour based on typical video data rates, at 12 hours a day, a 10 year old human has processed 43 terabytes of information
We are still far from uncovering the full power of the brain, but I can see this point being true in some aspects. We process a lot of information, but we don't have all that information stored in our brain, What we do have of it, most is in a sort of archived state and we'd need the right prompting to even bring it up (and like ChatGPT, we can indeed "hallucinate" with imperfect memories as we try to recall). So the brain's short term memory is pretty efficient in keeping us focused and only recalling what is necessary for that context (Or not. some store information better than others).
But in other aspects the brain is indeed doing a LOT of heavy work. Similar to a computer, most of the processor isn't usually taken up from active programs in the foreground. There's so many subconcious operations maintaining our body, and actions we optimize to take very little thought to execute (walking being the most common one, something most of us probably spent 2-3 years as toddlers striving to master). So you can make an argument that we are indeed supercomputers in some regards.
But that was all tangential. I think the main sentiment of that phrasing was simply that "we don't/can't memorize massive tables of information and recall it on the fly". Computers are extremely good at this, but are awful (relatively) at interpreting such information. Human minds are limited in processing raw information, but tend to be much better at aspects like pattern recognition, identifying subtle outliers, and connecting/correlating it to other types of data.
We give context to data, computers (even LLM's as of today) are simply attempting to imitate this with brute force of data (but slightly more elegant). In some ways it's simply because we use more than our eyes to recall and retain information; a computer can't "feel" heat on a smooth metal surface, associate it with a slight burning smell, hear the sizzle of smoke, and see a slowing red surface as a reference when trying to recreate the depiction of a stove. It can probably generate a nice looking stove regardless, but that lack of context can throw the rest of the scene off when it comes to contextualizing what else should be around a stove.
-
Comment on 'The gold rush is over:' Slay the Spire and Darkest Dungeon devs say that big Game Pass and Epic exclusive deals have dried up for indie devs in ~games
raze2012 Link ParentMorals aside: if an "inidie" game has to take out a large loan, it seems to challenge the sentiment many consider when they hear "indie". Is going to a loan shark or re-financing your house really...Morals aside: if an "inidie" game has to take out a large loan, it seems to challenge the sentiment many consider when they hear "indie". Is going to a loan shark or re-financing your house really that much different to finding an investor/publisher in that regard? It means either way that the costs for development are beyond a few people's labor out of the back of a garage.
-
Comment on 'The gold rush is over:' Slay the Spire and Darkest Dungeon devs say that big Game Pass and Epic exclusive deals have dried up for indie devs in ~games
raze2012 Link ParentYeah, because many indies of yesteryear nearly killed themselves trying to get their games out, and survivor bias shows us the victors. The Cuphead devs took a 2nd mortgage out on their house to...There's a reason indie exists in the first place
Yeah, because many indies of yesteryear nearly killed themselves trying to get their games out, and survivor bias shows us the victors. The Cuphead devs took a 2nd mortgage out on their house to get it over the finish line. Barone quit his job and spent years working 50+ hour work weeks to get out Stardew Valley (pretty much the classic bad decision many indies make). And these are just a few of the public accounts we know of.
There is a reason indie exists in the first place, and I don't think the social capital should be understated from the old ways, for what were technically simpler games in a less saturated market. It's a creative endeavor, and many creatives can definitely work themselves unheathily for "the sake of art".
This seems to be a discipline thing rather than a failsafe, so I think the bios would be a good first step. If they are getting to the point where they reboot their computer just to break their habit, it can escalate to hardware removal.
If it escalates to a point where OP is going out to a store and spending money to workaround the hardware failure, there are still solutions to prevent this. But they may need a stronger intervention than the internet can give them. You'd need a trusted individual to hold the BIOS password and disable the laptop's network card AND USB ports.