My engineers write better code when we enforce types.

It's easier to do this then retrain everyone on Go and rewrite all our code.

New stuff is often in Go now, but prototyping quickly in Python and then enforcing types when we have to get it ready for production has been working decently

A reverse funnel

Weren't we promised that quantum dead reckoning was right around the corner?

"works" and "fits in phone" are to VERY different levels of maturity of the tech

Asking it why it did something will not get you usable data. It will read what it did in the context and hallucinate a reason.

You are right on that, but my point was how it just ignores what it read, even if it is in its context

It read the context and randomly picked a way around what it asked. That's just how it works. Other times, it will follow it.

Porque no los dos?

Liquid seems like a better approach from an engineering standpoint because it is non compressible. But then I imagine dealing with liquid is more of a pain than air.

Someone said GitHub is racing to the mythical "zero nines of availability" and I love it

Hmm... 88.8888888%?

Jesus, that's both horrible and seems within reach.

They've already been well below that over the last 90 days

Yep, they just need to improve their reliability by 2%!

https://mrshu.github.io/github-statuses/

This page tells a very different story from GitHub own status page. What is different here?

Github measures/reports the SLA of the individual services.

The external page linked above goes the other extreme and considers it a bad status whenever any individual service is degraded.

In reality the majority of people only use 3 or 4 of the core services the majority of the time but since there's no "core services" SLA/uptime the usability of github for the majority of people is slightly obfuscated.

Part of it is that it considers downtime in any of the services GitHub provides as GitHub being down. So if GitHub had 100 different services, and only one of them was down at any given time (but at least one was always down), then it would show 0% uptime.

Tl;Dr blame Gatorade marketing.

The book "waterlogged" does a good takedown of the myth. Basically only you need to drink when you are thirsty.

Americans don't eat vegetables though

They also eat plenty of salty processed foods.

Which helps with water retention, so...

On a task by task basis the code Claude generates is pretty good these days. The biggest issue I see is that it wants to rearchitect the code constantly and I have no faith in my tests anymore because Claude will just "fix" them

I think some tests should be considered to be part of the specification rather than the product.