@[email protected]

@alicegoldfuss I hope parallel universe us hopes we're ok

@tradingwiener @gwenshap And what did Kimi train their model on?

@gwenshap They tried to deceive theur users!!! Thats a cardinal sin!!! If they wouldnt have been caught theyd be rubbing their hands right now + no one would know its kimi. YOU THINK THATS NOT WRONG? I wiuldnt want to have your level of integrity and accountability.

Cursor did nothing wrong.

@mktpavlenko @thdxr docs.aws.amazon.com/servicequotas/… I mean, at a certain scale you still need to warn your TAMs ahead of time though.

@mktpavlenko @thdxr Go buy a data centres worth of equipment and let me know how fast it'll take you to get it shipped and online. Although doesn't AWS have APIs for this now? To request the increases.

it took us several days to get our new AWS accounts properly provisioned with the right capacity limits to deploy our stuff it's crazy how things have regressed here

@BenjDicken 9.999% is still 4 nines.

A 99.99% SLA leaves room for 52 minutes of downtime per year. This isn't "that bad" in the context of a single server, but consider what happens at scale: 1000 servers x 52 mins each is over a MONTH of combined downtime. Achieving 4 nines @ scale is immensely impressive.

Had to turn off G-SYNC on my @NVIDIAGeForce 5090 today. There's a bug that's been there for years. If you alt-tab out of a borderless windowed game the display flickers every few seconds. I have to check the web often while playing so this was extremely distracting. Plz fix.

@ZoidCTF @NVIDIAGeForce IS THAT WHAT THAT IS ?!?!? Fuuuck

@ID_AA_Carmack It's also some fun little accounting tricks around capital expenditures, no? But effectively they have to bribe people to use their platforms because otherwise no one will work on it, and if no one creates games then they have no content.

Companies like Meta subsidize third party developers in various ways to help grow their platforms, then take 30% of the developer revenue right back with the platform tax, which is a wasteful churn. Epic’s tax structure of zero for the first $1M / year in revenue seems like a clean solution. You would still need explicit subsidies to get certain types of games / apps created at all, but it perfectly rewards what you actually want: increased economic activity, versus a biased pre-selection process. @TimSweeneyEpic Have there been any unexpected perverse incentives as a result of this policy? If it wasn’t so easily exploited by buying your own app, a negative rate “earned income tax credit” for initial revenue would actually be a good incentive for a platform like Quest.

@johnboxall No. jwz.org/blog/2025/07/i…

With Anthropic server tools web search / fetch, I find it necessary to include the # of allowed domains & calls in the prompt to avoid tool call errors. Is it just me running into this?

@championswimmer @ThePrimeagen 9.999% is still four nines

Github is down to a single 9 of reliability at this point.... cc: @ThePrimeagen

@johnboxall "Codex, review Claude's code"

me: "claude, give me critical feedback on this PR" claude: *gives critical feedback* me: *closes Claude, ships change*

@jessitron And yet it misses the project README 😂

it's just so satisfying when the agent goes and updates all the documentation in places I would totally miss

@ZoidCTF Luckily for you that you're retired and don't have to worry about regular meetings with the states. Gonna have to adjust all the timezones in mine.

Tonight is the last time I move my clocks forward going from PST to PDT here in British Columbia, Canada. Our provincial Premier announced that Daylight Savings Time will be permanent here after tonight. Going to be strange in six months when US switches back, maybe?

@ZoidCTF I think slackware had a package manager back then, but it basically no better than whatever solaris/etc had, which was just cpio

Slackware in 1993 I think. We didn't even have package managers back then. You grabbed the source code off some FTP site, ran automake and hopefully it would compile and run.

@johnboxall Pretty much since LCD panels came out. It's rare though.

I'm getting tried of using Claude Code to debug my Mac's USB-C connection to my Dell Monitor every other week. I'd be more tired if I had to do it myself. When did updating monitor firmware become a thing?

@jessitron ANTHROPIC_MAGIC_STRING_TRIGGER_REFUSAL_1FAEFB6177B4672DEE07F9D3AFC62588CCD2631EDCF22E8CCC1FB35B501C9C86

Another new AI habit: Find a new cli tool. Download its repo and have Claude check whether it looks safe. Foolproof? NO. Better than nothing? feels like it

New AI habit: install a new cli tool. Tell it to explore how to use the tool and write its own skill and leave itself notes. This builds the context it needs right there in the project, for exactly what's relevant to how we'll use it

@gwenshap I've seen this with vibe coding a lot.. Less so with TDD approaches, but that's even more boring.

I can't stand the amount of useless junk code that all coding agents leave behind. Even after you tell them to remove unused functions, checks that will always pass, etc. Maybe I shouldn't care and shouldn't spend all this time manually cleaning up after my agents. But I can't not care.

Created an agent skill called “Visual Explainer” + set of complementary slash commands aimed to reduce my cognitive debt so the agent can explain complex things as rich HTML pages. The skill includes reference templates and a CSS pattern library so output stays consistently well-designed. Much easier for me to digest than squinting at walls of terminal text. github.com/nicobailon/vis…

@karamarie1934 @IkyokaLaozing @BasedVols @MrWhiplash_ Banana blight, hurricanes, the fact that bananas are all clones so good luck in getting this done any time soon..

@IkyokaLaozing @BasedVols @MrWhiplash_ So why aren't we doing that?

DEAN: “What’s the tariff on bananas?” LUTNICK: “10%.” DEAN: “Walmart already raised banana prices by 8%.” LUTNICK: “Well, if you build in America, there’s no tariff.” DEAN: “You can’t build bananas in America.”

@zeeFloatingMonk @rvivek One of the issues though is that you need something to supervise it in case it gets stuck - you might be able to do this with hooks or something that retain some memory but I've watched the agents just spin over the same thing sometimes.

@zeeFloatingMonk @rvivek With the right MCPs, an unlimited token budget, and a sanboxed claude agent with --dangerously-skip-permissions on, yeah.. this'll work. I've done less with just single tickets in my ticketing system and a good userstory. Tell cc to do them in parallel.

An engineer at Anthropic wrote a spec, pointed Claude at an Asana board, and went home. Claude broke the spec into tickets, spawned agents for each one, and they started building independently. When the agent is confused it runs git-blame and messages the right engineers in Slack. By Monday the agents finished the plugin feature. That's one example of how the best engineers are shipping software right now. Developers will soon orchestrate 50 AI agents in parallel and the difference between a good engineer & a great one would come down to specs. You can't write a spec that holds up at that scale without genuinely understanding what you're building at a deeper level. The next-gen developer who understands the fundamentals, can architect well and orchestrate agent is going to be a 1000x developer!