apothegm

The ceiling fan in my bedroom was installed no later than 1981 (possibly a few decades earlier) and is utterly silent. On high you can hear the rush of air and thats it.

Of course they won’t tell us which ones didn’t…

What natural selection favors depends on the environment. That’s why dinoasaurs thrived and then went extinct, and why fish are different from dogs.

Yes.

If for no other reason than that our code can have major impacts on people’s real lives. We’re not developing games or something.

Ah.

I feel free to curse at it precisely because I’m confident it doesn’t have feelings, and it’s like swearing out loud at an inanimate object like when a shoddy screwdriver breaks, or strips a screw, or whatever.

Most people who write novels do so because there’s something (a story, an idea) that they want to communicate to other humans. Sure, you can write a novel for purely commercial purposes, but most people who write do so for the love of storytelling. Which is an inherently social and interpersonal activity.

Yes!

Declarative code has low cyclomatic complexity per unit of business logic. It also tends to have low state management demands (which the functional devotees will tell you is the root of all evil).

You can use a DSL or other abstraction so that you can declare what will happen in what circumstances, in a way that’s succinct and easily comprehensible. And all the repeatable loops and conditionals that make the declared thing happen in the appropriate circumstances are under the hood and needn’t be concern or be loaded into mental context by the person designing or updating the requirements.

It’s not the right solution to every problem, but when you can apply it effectively it’s a huge win.

However, agents and most SWEs are trained on imperative (wrapped in procedural/functional/OOP paradigms), and most will default to imperative because it’s familiar.

As I get deeper in my career I find myself gravitating toward a style that’s declarative where possible, falling back on functional, and then only when necessary imperative/procedural/OOP. It’s nice that many modern languages allow you to mix and match features of all those paradigms.

TL;DR: Mental load is the bane of programmers everywhere, and declarative programming minimizes mental load.

Fuck yes, I do. Not the most effective way to get things done, but a bit cathartic. And no, the stochastic parrot does not have emotions. And yes, it sometimes does worse after being cursed at, but you’re generally better off forking/rolling back the convo to a point before it screwed up anyway. And other times strong language manages to provide the emphasis you need to get it to actually pay attention.

Anyway, why are you ashamed of it?

Thank you, helpful context for those of us from elsewhere!

Perhaps for the particular image you liked. But choosing something “typical” or average over what it was instructed to do is a massively common failure mode for AI. One that makes the difference between a useful model and one that makes you want to throw your laptop out a window.

Yes, obviously. But historically that was the case, so it’s what patients mostly expect. Hence the loyalty after the sale, and the value to PE.

They sell because it’s their retirement nest egg.

Most private practices are small groups, not individuals. It absolutely sucks to have to find a new doctor who’s a good fit for you, when you’ve been seeing the same person for years or even decades. Usually when your doctor retires you rely on them to make a recommendation for a replacement. Usually that’s another partner in their same practice. At the very least, you (historically) expect them to sell the practice to someone they think is a worthy successor. Patients are fairly sticky.

And many of these PE sales require the selling partner to stay on for at least a year or two to ease the patients through the transition and so the patients get used to the changes before the doctor actually retires. That way, sticking around after the doctor retires doesn’t feel like such an upheaval.

Database Design for Mere Mortals. The first afternoon I spent reading and puzzling through that book was worth an entire semester of computer science classes on the topic — and caused me to fall in love with relational databases.

With special mentions to K&R as well as The Manager’s Path.

Pretty sure a 64-bit integer gets us to the death of our sun with several powers of 10 left over.

Sol won’t ever go nova, though, so I’m not sure how one would calculate the number of seconds until that happens.

Y2038, meanwhile, is just an artifact of the limitations of older hardware.

My dad gave me his password manager master password and his laptop password. I added both to my password manager. He tracks my mom’s most important passwords in the password manager, too.

We also talked about setting up a family account and sharing in his password manager, but it hasn’t become a priority yet vs the other late life planning stuff.

Still leaves an absurd amount of wealth to manage.

How would this be selected for?

The “good” news…

What does the existence of an unlimited plan have to do with whether people want or need your service badly enough to pay for it at all?

Good news for those horrified by the font: reader mode seems to work just fine, fwiw.

There’s so much more information to consume than news — or worse, tech news. Every day I find new things to learn about physics, history, geology, paleontology, biology… and not just things we discovered today, but decades and centuries worth of findings.

There is so much more to learn than any one person could even skim in their lifetime, even if you ignore absolutely all news. And heck yeah, it’s addictive.

Positive if you consider the disappearance of the arctic ice caps to be a positive thing, I suppose?

Yes that’s what it means for it’s output to be stochastic.

I switched from MP to Homebrew over a decade ago. At the time, Homebrew was less finicky and better at resolving dependency conflicts. It was a lot easier to get a system running MacPorts into a an irretrievably borked state. Not long after, Homebrew install instructions became ubiquitous, whereas with MacPorts it seemed like you always had to figure it out for yourself — if the package was even available, which it more often wasn’t.

Haven’t used MacPorts since. Has it gotten any better at those things?

Data-harvesting makes it sound far more innocent than it is. It exists to influence the outcome of elections. Through using the harvested data for propaganda, intimidation, and I’m sure they’re looking for ways to interfere with the actual registration rolls.

It is not reassuring that these things have a human driver, when that driver has no rear view window.

The LED will display it first, but will the signal reach your visual centers first? And how about the time for your brain to process that signal and make you consciously aware of the visual vs. the way the pain signal is forwarded to conscious processes?

If it’s allowed to write any sort of scripts, it will find a way to execute arbitrary code. If it’s got access to any form of internet connection it will find a way to use that arbitrary code to do as it wishes with that connection. So much for sandboxes.

And that’s setting aside “sandboxes” that are just instruction-based restrictions on what executables may be called. Anything that isn’t a deterministic external filter _will_ be ignored at some point.

Many though not all of these patterns are useful even in a small company.

I’m on a tiny team building an application that’s unlikely to ever average more than a handful or two of concurrent active users.

We’re using DB replicas for redundancy and failover and may eventually use the read replica to remove load off the primary. We’re using caching (which comes almost “for free” with our framework) to save slow/resource-intensive expensive DB and third-party API queries for lookups against rarely changing data sets.

We run background jobs and need at-least-once processing, which means consumers must be idempotent; not to mention idempotent handling of webhooks from one of our vendors. In fact, I can’t think of a small company I’ve worked at beyond proof of concept stage that hasn’t needed idempotence and benefitted from caching some things.

My last small company dealt with high volume and needed rate limiting and outboxes. In other highish volume contexts I’ve pre-computed expensive data structures (idempotently) and cached them in a document database.

Next time you’re at a small company might be a time to look around and ask yourself if any of these patterns is needed or would improve your software’s reliability or performance.

You start over from a point in the chat history from before they screwed up, or you start a new chat.