0o_MrPatrick_o0

Apple perceives itself as a prestige brand. That perception depends upon consistently high execution and a perception that they’re trying to create beautiful experiences for users.

They want the device to be a mechanism for you to purchase things. Hence the “Apple card” and all the hullabaloo around that. You’ll only use your phone if you perceive it to be secure. Apple’s investments in Secure Enclave’s, sandboxing are all directly linked to their belief that this will help the consumer perceive the safety of the product- and again, it goes back to the perception of the brand.

Quality craftsmanship has properties that outsiders can perceive- even if they don’t understand it. Apple makes those investments.

Google- well, they just make things and then abandon them. Android’s bizzaro world open source is a chaotic mess that reflects the lack of discipline in Google’s leadership.

The education profession in the United States has become a black hole that gives people who don’t plan the illusion of a career.

I don’t think this is a societal problem- it’s that people with the endurance for a “career” in this field succeed through the establishment of co-dependent rat-kings for organizational structure.

Yes.

I have to observe:

Seeing the raw CoT is the best signal for evaluating prompt efficiency. The fact that frontier providers don’t share this feels criminal after this last weekend. They’ve played us all for fools.

Thanks for the token setting observation. I’m running ollama headless and from the command line. I didn’t even know there was a web ui. Conf file is pretty straightforward though.

I settled on 64k tokens because of a recommendation from an LLM. I now see that that guidance is associated with similar sounding (but not the same) configuration constraints. So this weekend I’ll try turning it to 11 (250k tokens?). Anticipating some hard locks over here.

Thank you. This is the hackernews I miss.

FWIW, I’m not dead ended yet.

Using frontier providers to manage prompt tuning experiments has been very exciting. Might be something there.

Thanks for the feedback. I wanted to get into more detail, but I spent the whole weekend working these problems and then constructing this post.

Dario’s behavior this weekend made me feel like this just needed to get out quick. In the future, I’ll be sharing more details about some other things in the process and some ways I found to use automation to accelerate splitting prompts for use on local inference.

I mean this respectfully:

I am no longer and will never again be interested in Bill Gates’ opinion about anything. He needs to retire. He’s a sclerotic representative of what needs to be left behind.

Thank you for your contributions, Bill. You’ve done enough.

"complete wheel-spinning nonsense that was OpenRAN" is a concise summary of 3-5 years of cargo-cult telco innovation.

Hi Telco nerds!

It's looking like 3GPP is shifting away from faster networks and into networks that are programmable, energy-budgeted platforms.

Anybody doing experiments with components where power is transmitted by a transport provider? Ambient IoT sounds like a classic 3GPP proposal that utterly fails because every time they publish, nobody follows on with legitimate investment. Am I too cynical?

The tech world needs good heels! His worst takes were some of his best!

Dvorak's curmudgeonly approach to tech journalism was deeply enjoyable. He was utterly un-sycophantic and a playful troublemaker. Condolences to his wife, family and fellow podcasters.