esmevane, sorry

esmevane, sorry

@esmevane.com

Insipid but well-meaning software developing art gremlin. "Incoherent; mayhem." - 4/5 stars. "Couldn't make any sense of any of it, it's like he's talking to himself." - 2/5 stars.

250 followers355 following2.7K posts

I think it’s because people don’t internalize “thing I do to check that it works” as a test, honestly.

Modern Software Engineering@modernswe.bsky.social

"TDD is actually easier than writing the tests afterwards, so why is it so hard to learn?" - @emilybache.com What are your thoughts? 👇

this graphic is actually where i wound up on memory systems. just tell the robot to use a local wiki in the repo, it works really well.

Zed@zed.dev

We tested 25 approaches to MCP tool search, including decision models, LLMs, and embeddings. Our benchmark runs 150 search queries against 536 tools. We settled on weighted BM25 with a few normalization rules. Fast, local, no model or API calls.

Half expecting this guy to reveal a book deal or something coming down the pipeline and pull the "I'm just doing it for the engagement" crap like the Tomato Millionaire guy.

esmevane, sorry@esmevane.com

Even seen people doing the "ah but I have you (smug)" support of this fella as though a median analysis of a set is proper because it offsets extreme values of the set. No, those are outliers. The median changes when you remove outliers, doesn't it?

Even seen people doing the "ah but I have you (smug)" support of this fella as though a median analysis of a set is proper because it offsets extreme values of the set. No, those are outliers. The median changes when you remove outliers, doesn't it?

esmevane, sorry@esmevane.com

When you’re deciding what to do based on analytics, remember this kind of thing. It may seem to you as though you have an understanding of your data, and how to navigate it, but if that’s led you to make extraordinary claims like this, then you do not. You still need the extraordinary proof.

These as-a-service chatbots can't last. I mean, we're already in a world where arbitrary systems can now talk. Yet, we're talking to them via a chatbot on a remote computer that then talks to our computer, then reads the compiler output, then back. Why not just talk to the compiler?

When you’re deciding what to do based on analytics, remember this kind of thing. It may seem to you as though you have an understanding of your data, and how to navigate it, but if that’s led you to make extraordinary claims like this, then you do not. You still need the extraordinary proof.

Matt Darling@besttrousers.bsky.social

"but it seems pretty obviously true that it would tip some people from financially stable to not" It may *seem* that way, but this is not what we observe.

I don’t think Gruber would have pressed publish on his Guaranteed To Be Relevant To All Social Networks post if he wasn’t sure it’d drive tons of traffic to his blog where he Writes About Brands No One is Interested In Reading About Presently.

My amazing partner, who’s been yearning to watch stuff with me, minutes after I start the stupidest trash comfort movie I’ve got: “oh hey what’cha watchin’?”

Folks have long thought that executing on an idea was about as much work as perceiving it. There’s this notion that using robots changes it so it is. But really, what’s going on is that now you have to perceive the idea over and over again for the same amount of time.

Just grabbing one idea and holding it and forcing a robot to articulate it not just in rough form but exact precision, it’s still iterative, it’s still labor, and it’s genuinely exhausting to do as well.

There’s an upshot in the ramp up, and the scope, and the versatility, of what an individual can accomplish. But the dynamic is only hands off if you can afford not to care. If you have to care, you have to fully plunge in. So, it isn’t casual.

Totally! Good advice. My preference is to have a vague idea of: - major motivational plot beat - general appearance & personality - a driving goal And I try to 7-words-or-less whatever I can (Also I try to find a way to quickly accept & get along with other player characters)

This post is unavailable.

An old tension between me and a friend is “quadrafecta” (me; correct) and “fourfecteca” (him; Terry; wrong)

Dave Rupert@davatron5000.bsky.social

When it comes to web stuff… It’s one thing to have an idea. It’s another to be able to build the idea. It’s another to be able to explain the idea with words. It’s another to be able to build examples that illustrate the idea and the problem it solves. That’s quadfecta is difficult and rare.

View full thread

So, again: worrying charts like the linked graphics are not people gloating and being wrong about AI, or missing the point. They're an eyeball on vulture capitalists in the act, harvesting the middle / lower class yet again. It's not reckless to them, it's reckless (and predatory!) to us.

It's important to know that you don't have to be a master of economics to understand that the decades of decline we've seen all followed this exact pattern, and I'm very sorry to see so many folks, young and old, yet again not take it seriously.

View full thread

A bubble popping isn't a threat to the majority of these orgs because they don't plan to be the ones falling. They're planning to be poised to scoop up the spoils of whoever DOES fail, though. And the resulting bank run / harvest / bailout will be a huge, zero-consequence windfall for those left

So the main risk isn't "will AI stick around?" (it will) or "will we find useful application of it?" (we have and will). It's this: When one of the players in this phony money investment circle capsizes, what damage is it going to cause to _us_? Right now it seems like: a lot.

"Well, the younguns," I guess? It feels like if people witnessed the economic apocalypses we've had first-hand, we'd be more concerned? Then again, maybe not! Folks my age / older are ignoring it right now too. But in retrospect, we're all like: shit that sucked. And that's what bubbles do.

esmevane, sorry reposted

Finally gave the Advanced React website a fresh new look, aligned with the blog. 🎨 Who's here still reading actual books? Feels like a good excuse for a discount on some advanced React knowledge 😉 30% off for the next two weeks 💸 www.advanced-react.com

advanced-react.comAdvanced React by Nadia Makarevich: Deep Dives, Patterns & PerformanceThe advanced React book. 16 chapters and 105 interactive examples on how React really works, from re-renders and reconciliation to refs and performance. 4.8★ on Amazon, 4,500+ developers.

The strategy behind behavior like this is often to just keep asking and asking, preferably around different company each time, until you finally get enough people saying what you want to hear. It’s not asking for an answer, it’s looking for permission

greenlightening.bsky.social@greenlightening.bsky.social

We are living through ppl asking the questions, “do security and accuracy really matter, though?” as if the answers to both aren’t glaringly obvious.

Heavens grant me the self confidence of some of these dudes reposting their own not good jokes 3-4 times a day for days after making them. Yearning for that algorithm to boost your junk, boys?

The old Spotify algorithm kicked ass but it was tough to optimize and involved a lot of human parts. Here's my assertion: any recommendation system which a human will get value out of, long-term, must be curated by humans.

Time and time again, recommendation algorithms biff it. And it's easy to understand why: you can't both optimize for safety and surprise/delight people. Surprise is about being in front of something and it isn't safe. Optimizing for "good recs" is optimizing away from _enjoyable_ and _useful_ recs.

Or in other words, recommenders pick safe recs. That's also why Spotify sucks so hard these days - at one point there was a team of data nerds who tagged and linked songs with stuff only humans would. They're all gone, replaced by automation.

So that means that the old algorithms that would serve you things like Discover Weekly, and the year in review feature, the actual strange human tastes that guided those into the realm of actual utility have been replaced by an autocomplete robot.

I'm not criticizing the films or vanilla (it's my favorite flavor), it's not derogatory. Vanilla's an incredibly exotic thing. Extracting it is a culinary milestone. BUT - it's very safe. You can put it in most sweet things and it works. So, it's easy to recommend.

The problem here is that "easy to recommend" means that it's inoffensive, not that it's ideal. And that's why recommendation algorithms suck - they're primed to fulfill success criteria, and that means they hedge against failure.

Basically, Rotten Tom's Tomatometer stopped "working" for me at one point. I watched like 3+ 100% movies in a row over a year and they were all boring as hell, for me. Sure enough though, the flicks were highly regarded. So, what was going on?

I eventually realized it was this: the films weren't great. They were safe. No critic watched them and thought, "wow, I can't recommend this to people". They were vanilla.

One big mistake I see with design token systems is that they attempt to make things scale predictably, using the golden ratio usually. That _seems_ to work, but it just doesn't hold up. You have to fine tune the ratios for every given setup.

well, this morning the robot reviewed one of my old codebases and called the git log "a journal of someone building alone" so, i believe i should be allowed to take a personal day today and go back to bed