https://www.youtube.com/watch?v=C33-YCX7Too

In ice none of these really stop overcorrection, or at least they don't in my 2020 truck on icy hill/mountain roads in Maine. And I've seen nice recent Volvos and BMWs with presumably the best safety tech in ditches up in the ski towns. The correct safe speed to drive on icy roads is not to drive at all of course, but people have to get places and people make mistakes. IME the assistive technology defaults don't do great on ice roads on some kind of up/down grade.

AFAIK drivers can still steer and brake themselves into a loss of control situation on ice regardless of safety features. So I guess I'm hoping once you take those two variables out of their hands, the FSD vehicles will be safer. Who knows though.

I went many years without a loss of control and the one time it did happen (logging roads with ice pack) was enough for me to buy Nokian studded winter tires to minimize the effect of ice as much as possible.

And if the car reduces speed when appropriate and some assholes start tailgating it, it won't suffer the anxiety of holding up 10 cars that want to drive beyond the safe, reasonable speed for the snowy/icy conditions.

It's also caused an uptick in inbound to dev tooling and CI teams since AI can break things in strange ways since it lacks common sense.

I'm not here to defend brodozers, but you cannot possibly prove this statement. That a _pickup truck_ isn't hauling the majority of the time it is on the road is not some new thing. But of course there are more pickup trucks on the road than ever, so if you argument is aggregate time of all pickup trucks not doing truck things is the highest its ever been is certainly true, but you'd probably have to go back to before the 80s for that number to actually be meaningfully different per truck.

That said Medium did pay higher profile writers and publications to move to their platform (in some cases quite a bit of money) in a similar way that Substack has, which was to dip into the VC funded bank account.

* Patreon kind of sucks for writing / newsletters or else they could have captured part of this market. Medium was supposed to fix this but they had an epic collapse.

* Single point where subscribers can manage their subscriptions and preserve a common identity across subscriptions in the comments etc. Again, Medium/collapse.

* A rapid adoption of substack by well known online writers with loyal followings. These writers either had their own blogs or were exiting traditional media or getting dumped out by the collapse of online media (gawker network, buzzfeed news, etc). Again, could have been Medium if not for their collapse.

* I suspect Substack spent a lot of that VC money guaranteeing 2 years of X revenue for a non-trivial number of high profile writers so they would onboard.

Reading this story I didn't realize just how much they had taken over the years (I use to operate in this media space, but haven't in a long time). I'm not sure what the headcount is but that amount of money is staggering and I can only imagine it was all used to acquire DAUs and very little novel technology has been created with it.

I think Substacks first 10m/100m (their keep of rev/total rev from subscriptions) was extremely impressive and fast. But also it was a kind of low hanging fruit. There was a market already there for this and Medium/Patreon couldn't capture it. Now if they are really at 45m/450m that is much less impressive. It will be extremely hard to get to 100m/1000m and IMO impossible to get much higher than that with their current approach.

All `kubectl` does is create REST API calls to the control plane. So to be fair, what I'm grousing about can be accomplished just fine by a developer like me constructing API calls from python to update objects in k8s.

The problem is I work in devops where tooling written in proper languages with standard libraries that have things like useful arrays or robust string manipulation or ergonomic concurrency is a non-starter for some reason. The argument against that being mostly "I have to install the interpereter first".

That's a totally fine trade off for actual sane array/list functionality, robust string manipulation etc. I'd rather form shell commands in a programming language than in bash. People seem to love it.

I do not think for a second though that the average person that "knows bash well" can read and comprehend a multi hundred line bash script written by someone else as fast or even correctly has a seasoned python dev reading python written by someone else.

Even the helm infrastructure that I work in is completely wrapped in custom shell scripts that call all sorts of other commands to populate helm variables.

But yeah it's silly that helm templates require all sorts of {{ indent | 4 }} type incantations when the final YAML output is just sent through some kind of toJSON anyway.