743 karma · joined July 6, 2022
https://news.ycombinator.com/item?id=42698137
I'll make another prediction: their ultimate goal is to start charging a subscription service for Windows.
Or is the assumption that the training set is so big it doesn't matter?
I work at a billion-dollar EU company that’s balls deep in Azure after a very, very long migration away from on-prem datacenters.
Cutting off US-based cloud providers would be chaos of biblical proportions.
Step 2: make online login the default option
Step 3: make online login mandatory, leave an escape hatch for technical users
<we are here>
Step 4: remove the escape hatch
Sure, venv doesn't manage Python versions, but it's not that difficult to install the version you need system-wide and point your env to it. Multiple Python versions can coexist in your system without overriding the default one. On Ubuntu, the deadsnakes PPA is pretty useful if you need an old Python version that's not in the official repos.
In the rare case where you need better isolation (like if you have one fussy package that depends on specific system libs, looking at you tensorflow), Docker containers are the next best option.
UPDATE: 0.006%, not 6%
Short train connections are a gamble here, and worse still that risk isn't properly communicated to travelers as the official SNCF trip planner will happily schedule passengers on very tight transfer windows with no warning. I'm travelling across France soon and the planner put me on a a 9 minutes connection just before the longest leg of my trip, which is a 10h overnight train.
There's no way I'm going to gamble my entire holidays on a 9 min delay, and thankfully my ticket lets me take an earlier train for the 1st leg, but I pity the clueless tourists who blindly follow the planner and end up with their plans ruined.
But that does not help you much if every train for the day is already booked out.
I'm worried that Microsoft is succeeding at making the semi-proprietary version of VSCode the only viable option, while keeping the OSS forks as second-class citizens at best.
But humans can be held accountable, LLMs cannot.
If I pay a human expert to compile a report on something and they decide to randomly make up facts, that's malpractice and there could be serious consequences for them.
If I pay OpenAI to do the same thing and the model hallucinates nonsense, OpenAI can just shrug it and say "oh that's just a limitation of current LLMs".
Tip: you can click on the canvas to "disturb" the flies
I've been using it in a toy OS project and it works brilliantly.
That's a weird comparison. The goal of the space race was never to make money, and it was funded by the taxpayers, not VCs expecting a return on their investment.
The entire Xbox Series catalog, for starters. You can't even setup the console without phoning to Microsoft's servers. Meaning units not already associated with an account will become paperweights once Microsoft inevitably kills support.
But 14 of those were caused by the shuttle alone. All the others were over 50 years ago. So far, all the spacecrafts still in use today have had a pretty good track record.