conradev

In the few instances that I’ve been able to see a total stranger prompt a model, I learn a lot about them.

The model output is fairly uniform because the models are fairly uniform, but the input is how I understand the prompter’s theory of mind for the LLM.

The higher fidelity the theory of mind, the more productive the resulting conversation is. Basic things, like knowing what the model is even aware of.

It’s okay at subjective product judgements with limited context, and so the best engineers make the most important decisions themselves, constraining the model’s solution space to something looking like success.

Maintaining consistent progress towards a common goal with a bunch of different perspectives is the job.

I’ve seen plenty of humans manage other humans in this low effort way. It’s a bell curve, after all!

If a model does something wrong, I don't delete it and rephrase. I explain where it went wrong. I get better results this way!

I went to Amazon the other day and ordered a fully custom piece of software. I chatted with the seller about exactly what I wanted, what areas to focus on, and they ended up teaching me a thing or two about the domain. Very similar emotional experience to using a language model, which I’ve totally done before.

It is encrypted ciphertext rotated every 10m in lockstep with your Bluetooth identifier.

You will need to turn off your Bluetooth if you want to stop advertising a rotating random number.

The only difference with the contact tracing is that a key, on the device, can correlate the ciphertext. I have it off because the utility was never demonstrated.

Borderline useless? Is it just that the numbers are unit-less?

You can extract meaning from them! Find your current computer, find the target computer, calculate the percentage difference. 1000 is 1% better than 990, so it indeed doesn't mean anything at all.

The government does subsidize basic banking, though?

This is a good joke because they spent $2.4B on marketing in FY25

There’s this crazy thing that happens where you need to use carbon to decarbonize.

Solar panels are made in factories where energy and inputs are cheaper and then shipped on diesel cargo ships to the rest of the world.

Our cars run on gas, so we need new electric ones, but the ships and car carriers are themselves… not electric yet.

Because energy is more expensive, everything is more expensive, including electrification.

MPEG-TS is used to contain h264 chunks for HLS. MPEG-DASH and the new CMAF standard use fMP4 containers instead. My personal take is that Media over QUIC (MoQ) should support both.