mirth

@tinker@infosec.exchange @skykiss@sfba.social I attest that Tinker Secor's sage guidance helped my regulated-industry enterprise deliver record profits while maintaining compliance with relevant laws.

@WildEyedBoyFromFreecloud@masto.ai @GroupNebula563@mastodon.social @gwynnion@mastodon.social Ahistorical. There always have been some people who care, and there still are, but it's never been most people.

@gwynnion@mastodon.social I don't know who that sales pitch is for, truly the stuff of nightmares. A pitch that might appeal to working programmers is more like "can delete 25,000 lines of code every day".

@NanoRaptor@bitbang.social It is nearly certain that there will be a worm-driven Rick Rollening. Lurking in the background of scenery, together in every group photo, every camera roll will be a Rick Roll.

@NanoRaptor@bitbang.social Were the modules "Quadrants"?

@mnl@hachyderm.io I don't know of a single reference, Vitalik Buterin's blog read in reverse chronological order will get you pretty far. Many of the widely quoted authors in the area have the gross misunderstanding that they are solving societal problems with technology. Lots of interesting gizmos and gadgets, not a lot of actual solutions (or "utility" as they like to call it).

@tragivictoria@mastodon.com.pl @zoeyTheWitch@social.treehouse.systems @thomholwerda@exquisite.social Just the coordination of maintenance after such a unification were done would be its own complexity. Certain things, like conformance test suites, could be really helpful to share even if the underlying implementations were different.

@coldfish@sfba.social @cxiao@infosec.exchange I think a root enabler of many of these problems is too much corporate power over employees. If it was easy to quit many of these things wouldn't be workable.

@dank I can't tell if I should block you or buy you several beers, but either way I've clearly been on the internet too long.

@aparrish@friend.camp I've been part of various licensing discussions and one of the big worries when you're small is the "AWS problem": What happens when a big company decides to sell a competing service on your thing? Can you add friction and at least make them write their own? Should you have open code at all? None of the answers are all that great.

@azonenberg At first I couldn't tell if this was real or not, but then I realized that just by writing this you have made it real.

@mcmartin@mastodon.gamedev.place @TomF@mastodon.gamedev.place Yes Quake defaults were arrow keys and I switched to WASD after reading I think an actual paper magazine article that mentioned it.

@aparrish I do think OpenAI has about as much integrity as WorldCom, but also like WorldCom a lot of the stuff they're working on will still be around in some form in 20 years.

@mnl@hachyderm.io I know a good bit about this area, there's a lot of noise and not much grounding in the actual needs of any people.

@whyrlpool@slime.global Oh I want 10Gbps or higher, and so many devices already have that available in their USB interfaces these days just without an easy switch option.

@gregeganSF@mathstodon.xyz @antopatriarca@mathstodon.xyz @johncarlosbaez@mathstodon.xyz The magic (and part of the trouble) is the behavior is implemented as massaging points around in high dimensional space with implements with data-derived shapes. The human-visible primitive operations are more like "giant matrix multiplication" than anything we would recognized as part of a hand built algorithm.

@leo_wallentin@mastodon.social @Remittancegirl@mstdn.social As a long-time Linux user it does seem like recently the quality of life improvements counterbalance the sacrifices moving from Windows or Mac. There are still integration challenges but the overall usability is pretty good and simpler in a lot of ways.

@inthehands@hachyderm.io It's hard to look people in the eye and say your decision cost them the paycheck they use to feed their kids. Maybe that's still the right move? And how do you communicate your philosophy on conflict to leadership in other major countries, like China? It creates a lot of questions. You could well be right, but I'm not sure. Unfortunately power is a lot easier for total sociopaths who don't weigh the consequences of their actions.

@inthehands@hachyderm.io You and I would approve if Tim took a stand, and I think odds are good he could have stared down Trump, but it also could easily have backfired and resulted in a ban on Apple products. "These China phones, you know, they're all made in China, we can't have that, bad, very bad, and from now on we're only going to buy American." Apple itself has so much cash it will survive almost anything, but any protracted slowdown.

@inthehands@hachyderm.io I suspect what he's doing is hoping that as his parting gift to Apple he can function as the sin eater on their behalf to absorb the reputational cost of what he views as necessary political theater, giving the company better stability and the new CEO some room to distance themselves from this mess in a future where this gets mostly unwound.

@inthehands@hachyderm.io Tim personally isn't very vulnerable, he could get on a plane and live out forever in Switzerland or wherever he chose, but the rest of the company is effectively held hostage by random moronic trade policy. What he appears to have done is accept the indignity of being "Tim Apple" and presenting a token tribute that was the least gaudy yet gaudy bauble they could come up with, in the hopes that it would maintain stability. I'm not sure it's right, but I'm also not sure it's wrong.

@inthehands@hachyderm.io I half agree, but here's another perspective: Apple directly employs around 150k people with maybe 1M more working primarily on supporting Apple at suppliers etc. All of that is connected by a supply chain that is built on top of agreements and trust that Apple will be a reliable (if difficult) counterparty for the various companies and governments involved. Faced with a capricious and impulsive president, one who threatens all of those jobs, what is the right thing to do?

@aparrish Whether this stuff sticks around is of course the realm of conjecture, what I've seen technically both now and the trajectory makes me think there will be permanent shifts. OpenAI is unlike WorldCom in a few big ways, one of which is it had no pre-existing business. I'm not really sure what will happen to it but some mix of WorldCom and Netscape is my best guess. Name disappeared from view, the technology dispersed, well before the web got mainstream useful. We'll see...

@aparrish All of that activity gets built around some underlying idea about long term demand though, which (my opinion) is built on "build more & kill more faster & w/ fewer people in the loop" or at least the FOMO related to not being ready if other companies / countries get there. A16Z and Black Rock and OpenAI and Microsoft and the DoD and the FTC can all line up behind one plan.

@aparrish I think the key behind the incredible level of funding comes in two parts: first, the idea that it could be a direct replacement for types of human labor that have to this point resisted automation, and second, the idea that the technical capability imparts some kind of geopolitical advantage comparable to access to energy. Scammers and profiteers show up, of course, but I don't think they're the prime driver of this.

@kornel@mastodon.social Reviews have a lot of functions including quality control on the change, determining separate from quality whether the change fits the project's plans, disseminating knowledge to both parties about how the system works, building relationships between contributors, etc. I think part of the answer is we'll see more projects split into core + contrib where the design of the core contains the effects of bugs in contrib, and where the two projects have different contribution rules.

@jedbrown @jenniferplusplus The asymmetry is the core thing that concerns me. I can say that empirically starting somewhere last year LLM-assisted bug hunting started to be effective. The false positives are avoidable but the cost of remediation has not gone down with the cost of exploits. This new model may make the situation worse but we're already in it.

@MichaelTBacon @alienghic I've been trying to think of a happy ending to this story but I think it's going to be a tough ride.

@budududuroiu @jenniferplusplus I wouldn't give Anthropic's motives a lot of credit here but LLMs do make bug hunting much easier.

@MichaelTBacon @alienghic Do you have a source for that figure? I have access to some developer usage data for a customer and I don't see anything close to the token usage it would take to get there. Maybe $200 a month.

@mhoye@cosocial.ca Oh, it's still the CEO's fault and, although they may have been corrupted by a life of coddling without direct feedback from reality, customers know. Writing a blog post about a very poor post-mortem is worse than useless to their customers and company. None of this is new though, the venture backed startup world always has a slice of sheltered upbringings experiencing their first encounters with objective reality.

@MichaelTBacon @alienghic Training is necessary for better models, as models pass "good enough" for most tasks they're only necessary due to competitive pressure, and when a funding drought comes the competitive pressure gets released. Imagine a game of musical chairs where the "sitting down" part is who can cut costs the fastest. The management know this, they spend prolifically to force out competition while protecting unit economics to be positioned to survive the downturn.

@MichaelTBacon @alienghic I'm not sure I understand, what about the current cost is unsustainable? If to use some round numbers $100k of server hardware that costs about $40k a year to operate can serve subscribers paying a total of $7k a month then that's about a break even investment (assuming a 10% discount rate). Of course the big AI companies are spending a lot of money, but if they fire all the people outside a core product operations, engineering, and some research, the math is different.

@ge0rg@chaos.social If you find the wrong bugs sometimes you light the whole thing on fire to reduce spread.

@shana@mastodon.gamedev.place Nobody has pointed out the most important* difference, which is that one is very painful to write on a chalk board, while the other is very painful to write on a teletype. *well...

@mhoye In my work I can see into a few software organizations that are deploying this stuff. To varying degrees they're doing more, but it shows up as a bit higher product polish and completeness, fewer open jobs, faster design-to-ship times for features, more willingness to "eat the frog" and rework or remove stuff that's not necessary. Maybe a 2x to 3x. The benefits accrue to the shareholder and to some extent the customer, not as much the employees.

@monkeyborg @anne_twain @sashin Agreed, additionally it would be unusual to use the "Prof." title as opposed to "Dr.". The bio reads like a field from a database of prompts.

@alex@gotosocial.theonecurly.page @cwebber@social.coop I have some problems with this project but that's not one, doing a file-by-file module-by-module translation in place is a reasonable approach. Ideally with enough glue that new code can run in old tests and the reverse. Rust is a good choice for them, I'm skeptical that Zig is well suited for that type of project, though C++ or Go would also be fine. Was this good stewardship of their dev community, or natural resources? No.

@elizayer That's consistent with my experience. I suspect adoption will facilitate a divergence, where there will be de-skilling in some cases, while in others users will get the technical leverage to do more / larger experiments that they can then learn from. Tools to amplify differences in culture, temperament, available resources, etc.

@DaveMWilburn@infosec.exchange @mattblaze@federate.social But old fashioned arson means someone needs to show up in person at the scene of the fire, a special cable can be same-day fulfilled from a warehouse Amazon Prime style. Now, power-only USB-C cables...

@alienghic @MichaelTBacon @cwebber Not a lot of progress that I'm aware of. As you point out code/data separation is a foundational design approach used in normal computing but current LLMs don't really have an equivalent and I don't think most people working in the space are even aware of the idea.

@mawhrin@circumstances.run I have a hard time telling, maybe I'm sensitive because I'm often the subject of sincere arguments that the old thing/time was better when the underlying reason boils down to it being convenient because some underclass got to bear the cost of the externalities.

@violetmadder@kolektiva.social @inthehands@hachyderm.io As in other areas of life there are a variety of personalities. Corporate leadership suffers from some pretty strong adverse selection, but not all execs are sociopaths. Maybe half or less of those I've spent time with, not worse than operators of small retail businesses, or leadership in non-profits.

@whyrlpool@slime.global I would pay money if I could get that in a form that could pretend to be high speed Ethernet.

@gregeganSF@mathstodon.xyz @antopatriarca@mathstodon.xyz @johncarlosbaez@mathstodon.xyz Oh yes, I'm sure over time various structures within them will be understood and many replicated using more efficient and inspectable mechanisms. I don't think there's any fundamental reason the model weights have to contain the spelling of "Barcelona" or learned machinery to do trigonometry, Boolean logic, or sorting.

@mjd @nixCraft I have access to some developer usage data and it's not close to $5k per seat per month worth of usage. I don't doubt some organizations with bizarre incentives to use more resources can use much more, but ordinary developers for the most part don't seem to.