We already don't have free speech. There's nothing protecting it (and many laws already to the contrary.) There aren't really any such constitutional protections from what I can tell.

Once laws are passed they aren't revoked. So it's just a matter of political climate. Just wait for people to get a little more negative, a little more paranoid (which has historically been "helped along" in various ways)-- a law only needs to pass once, and then we're stuck with some stupid bullshit forever.

It doesn't really seem like how you'd want to design it.

>It takes care and careful engineering to produce good results. One must work to keep the models within the flight envelope. One has to carefully structure the problem, provide the right context and guidance, and give appropriate tools and a good environment. One must think about optimizing the context window; one must be aware of its limitations.

In other words, one has to lean into the exact opposite tendencies of those which generally make people reach for AI ;)

Hi HN! This is probably the most important thing I ever learned, so I'll repeat it here for the people who only read the comments :)

I found a way to work on my project every day, without willpower.

I just eliminated all the failure modes.

1. Skipping days breaks momentum: I can't skip days. (So I made it easy to win: I only need to show up for an hour.)

2. Delaying the work leads to missed days. I get tired, I get distracted, I forget. So I decided I need to work as soon as I wake up. Then I guarantee it gets done.

3. Distracted by stupid BS. Noticed that all my distraction came from the internet, and getting distracted would tank productivity for the remainder of the day. So I just unplug the router before bed.

That's my whole system, and I'm using it right now. (I used it to write this post!)

The deeper idea here is ... just find out what's making you fail, and see if you can't design around that. Just design it so those things do not happen. Then you will win by default.

Hope this helps! :)

---

P.S. the astute reader may observe that this describes the Q2 category in the Eisenhower Matrix... the stuff we want to get done "someday", but which will never get done unless we make time it. For this reason, I very strongly recommend dedicating the first hour to Q2. Q1 "handles itself", by definition. Q2 is the one that needs love!

(That probably should have been the main thesis. Oh well!)

Context: https://files.catbox.moe/eqg0b2.png

I think they later made a Black Mirror episode along these lines. "Resume viewing... Resume viewing..."

What are you using for transcription?

I tried Whisper, but it's slow and not great.

I tried the gpt audio models, but they're trained to refuse to transcribe things.

I tried Google's models and they were terrible.

I ended up using one of Mistral's models, which is alright and very fast except sometimes it will respond to the text instead of transcribing it.

So I'll occasionally end up with pages of LLM rambling pasted instead of the words I said!