I think the root problem lies in some of the really gnarly, pre-1.0 code that we are afraid to touch because it might break something, that’s why it keeps cropping up even though we keep slapping patches pell-mell on top of it. It just happened to pop out pretty bad in 03.
I don’t know, this whole situation could have been avoided if people would just check with the team before merging stuff. I think people see revisions from old releases so they just merge them because they “worked before” so they should work now. It doesn’t work for the current userbase, frankly it didn’t work all that well for most users back when it was released the first time. Just merging any old crap straight into main without reading it so they can brag about how many PRs they close.
Trying, I’m stuck on a bridge with some of the seniors right now. One of them wants to “just push through and see what happens” and roll forward with the change, the other one disapproves but clearly has no idea what else to do. Nobody seems to want to even propose a plan because then they’d be on the hook for it, so we just keep going back and forth and not. ending. the call.
yep, the bug triggers when a political party overflows its buffer. The funny thing is, the system still catches the error and spams the logs with FATAL error messages, but the system has been throwing them for so long and seems to still be chugging along so we kind of ignore them. One of the original devs made a comment about how someone should fix it right before leaving the team. Anyway, the effect tends to accumulate, so if you deploy it to prod and don’t fix it, eventually you get… this.
My bad.
Sorry everyone, I opened a new branch off the 2003 release while I was researching some of the bugs in the legacy code. That wasn’t supposed to get merged in…
This is in some ways an easier problem than classifying LLM vs non-LLM authorship. That only has two possible outcomes, and it’s pretty noisy because LLMs are trained to emulate the average human. Here, you can generate an agreement score based on language features per comment, and cluster the comments by how they disagree with the model. Comments that disagree in particular ways (never uses semicolons, claims to live in Canada, calls interlocutors “buddy”, writes run-on sentences, etc.) would be clustered together more tightly. The more comments two profiles have in the same cluster(s), the more confident the match becomes. I’m not saying this attack is novel or couldn’t be accomplished without an LLM, but it seems like a good fit for what LLMs actually do.
Why not? if LLMs are good at predicting mean outcomes for the next symbol in a string, and humans have idiosyncrasies that deviate from that mean in a predictable way, I don’t see why you couldn’t detect and correlate certain language features that map to a specific user. You could use things like word choice, punctuation, slang, common misspellings, sentence structure… For example, I started with a contradicting question, I used “idiosyncrasies”, I wrote “LLMs” without an apostrophe, “language features” is a term of art, as is “map” as a verb, etc. None of these are indicative on their own, but unless people are taking exceptional care to either hyper-normalize their style, or explicitly spiking their language with confounding elements, I don’t see why an LLM wouldn’t be useful for this kind of espionage.
I wonder if this will have a homogenizing effect on the anonymous web. It might become an accepted practice to communicate in a highly formalized style to make this kind of style fingerprinting harder.
Without knowing what posts they’re talking about, probably. I try not to drag US politics with me everywhere I go, but the fact that my country is ten years into the dumbest fascist coup in history and they still might succeed in knocking over our democracy is mentally invasive in a way that’s hard to put into words, and it tends to distort the lens through which one sees the world. Now that I reflect on it, I don’t think a week has gone by in the past decade where I haven’t thought about Trump, MAGA, and the mortal threat they pose to me and my friends. Maybe a few weeks in the beginning of the Biden admin, when I genuinely thought we had dodged the bullet and the Republican party would ditch Trump, but even then I knew the fascist problem hadn’t gone away completely and would be back. Its constant presence tends to make me see things through the lens of US politics even in places where it might be a reach. Combine that with the fact that there are just a lot of Americans with internet access and I get why it gets annoying.
Sorry y’all have to hear about it all the time, if it’s any consolation I too am tired of it and hope we can get our shit together, but it’s a tense time and probably will be for the next while.
The people yearn for IoC
Seems like there could be a crowdsourced version of Consumer Reports. A standardized battery of tests for each product category, and different youtubers could test products according to the test and produce (ideally reproducible) reports for each product. Not sure how the standardized tests would be created or maintained, or how the whole thing would be funded. But it would be cool to have some common, non-commercialized benchmarks that do what Consumer Reports does, but with better transparency and less opportunity to fudge.
It depends a lot on what you want to do and a little on what you’re used to. It’s some configuration overhead so it may not be worth the extra hassle if you’re only running a few services (and they don’t have dependency conflicts). IME once you pass a certain complexity level it becomes easier to run new services in containers, but if you’re not sure how they’d benefit your setup, you’re probably fine to not worry about it until it becomes a clear need.
It’s fun in a way that defies comparison.
As someone who was recently laid off if anyone wants to front the cash I’m currently available for cheap.
That’s why I have one host called theBarrel and it’s just 100 Chaos Monkeys and nothing else
They aren’t a waste of money if investors can assume that there won’t be party changes in the future.
When I went to buy fancy cologne for a wedding they had little bowls of coffee beans that were supposed to be palate cleansers. I cannot vouch for how well they worked, I felt like my nose was blown out after a few samples.
oh yeah if your engine timing is off it can make the whole system run really rough, even if it’s in otherwise superb condition. That throws a lot of newbies who don’t understand why none of their performance tuning seems to have any effect.
It’s related. All over the codebase different devs have put in catches for it, but often as not instead of handling it they just wrap it in a new warning, dump it to the log, and move on. Dwight wanted to decouple those modules, correctly, but he didn’t know how because the reason they’re coupled in the first place is because we didn’t handle this problem correctly in the initial design, so he slapped a warning on it and called it good enough. Every once in a while some hotshot dev comes in with a proposal that exploits it (and introduces tons of technical debt that we don’t see because we don’t measure it) and because it looks good for the shareholders the PR gets approved and merged. If they’re lucky then they get promoted out of development, or occasionally fired, but in either case the problem in the code gets exacerbated and never really fixed.