Kerbonut

This is the exact reason why the Apple Vision Pro is failing, because they put a toy OS on it instead of an actual operating system that is conducive for getting real work done.

Can you standardize with all of your tools on a common registry key and settings paradigm and same plugin/skills paradigm and common method for enterprise delivery? It’s a pain in the neck getting all of this set up and deployed throughout the org.

A fellow pyanodons player I see

They most likely truck in the brine from a mixing facility to reclaim the oil and truck out the brine to disposal when refilling the oil stock.

They look like above ground pits. Empties are showing the white paint and ones with brine look dark.

Wire typically means internet connection, and it’s hardly the bottleneck

Tailscale Aperture is having trouble doing the OAUth2 connection. Can you support oauth2_dcr or test with Aperture?

Claude Code is not the model, it's the harness. You can use any model you want with Claude Code to varying degrees of success. I use Qwen3.6-27b daily with Claude Code as an example.

> put a temperature sensor on the oscillator

At that point, couldn't we just use the temperature value to compensate for the drift?

Wouldn't a temperature compensating algorithm be just as effective?

Wrong. They disable it because it’s a safety thing. They still count it towards their metrics that FSD was enabled.

It’s still counted as FSD enabled. Would you prefer FSD to remain active and potentially cause further collateral damage after an accident causing who knows what kind of damage to the vehicle? Safer to shut it down when systems are working and brace for impact. Seriously use your brain.

That's like Joe Pesci in My Cousin Vinny where he stays overnight in a cabin and wakes up and freaks out at every little noise, but when he's held in contempt and there's a prison riot going on, he's sleeping like a baby.

I built my own tooling around templated manifest files (jinja2) and management via ansible playbooks (templated).

Almost like a "soft"-brick, if you would.

I had an idea somewhat related to this where we use the solar winds as a sort of road and the earth's magnetic field as a sort of rotor to convert kinetic energy from the sun into electricity.

"You need the latest version of the Google app or Chrome Chrome."

Looks like your units are in ms, so 0.20 ms.

Imagine a whole git repo of a project materializing like that.

Apparently, OpenAI API “credits” expire after a year. I stupidly put another $20 and trying to blow through them, 4.5 is the easiest way considering recent 4o has fallen out of favor for other models and I don’t want to just let them expire again. An expiry after only one year is asinine.

Is it just me or do they have some of the left/right images swapped?

I wonder how the LG appliances work for this. They also send data over sound for diagnostics.

The source is the training data and the code used to turn the training data _into_ the weights. Thus GP is correct, the weights are more akin to a binary from a traditional compiler.

That sounds fascinating. Would you mind writing up a demo on how to do that?

Do you mean that agents are being hyped in the same way data science was in the 2010s, or that they’ll have a similar impact over time? Would love to hear more of your thoughts.

I was thinking the job poster would have to put up the bulk of the pot, let's say a 20% "signing bonus" that is distributed amongst the finalists (job offered). Applicants would each stake a small amount each, which would be shared amongst the applications who are dropped out at each gate prior to job offered gate, say 20% of it is shared amongst the first round as an example. It would have to be structured to mitigate gaming the system on both sides, e.g. professional applicants who drop out after job is offered (never accepting a job).

Stop calling it a OTP and call it a temporary PIN :)

I wonder if a type of escrow service could help solve the problem of hiring managers getting flooded with low effort applications and simultaneously solve the problem of applicants being passed over when they are obviously the most qualified for someone else who looks good on paper but who is actually mediocre.

There could also be a case for some kind of ante that applicants have to contribute to when they apply. Pass the different levels of interview and you get a portion of the pot. Make it to the job acceptance and win the pot and if you accept the offer you get what the employer staked.

Maybe something like that could help solve this issue. Either way we definitely need more structure and better defined processes for both sides of the job hiring process (looking for a job as a prospective employee, and hiring to fill a position). It would be great if we could automate this in a way that is mutually beneficial to everyone involved and had more transparency in the process. Right now there is zero accountability on either side, and as TFA demonstrates, the balance of power has shifted towards the applicants recently.

I am right there with you on the open weight models. Only reason I started back on ChatGPT is because I'm rebuilding my servers and had to take out my 3090Ti. That thing is huge and I needed the room for hard drives until my 3.5" to 5.25" adapters come in.