Redster

What value to you get from having this happen on every turn with a change? I get that the ratio should be a lot higher than the default. Doesn't doing this for every turn where something changed slow things down a lot without gain? I haven't experimented, that's why I'm curious. Why not 1 line or 2 in a CLAUDE.md or AGENTS.md so that the agent ups the frequency, but doesn't make it every single time?

I've never used Iroh; can you use it with no setup or without an account? I send/receive files often enough from non-technical people, having a no-install/accountless solution for sending files is essential. For my use-case, my go to is croc or Bitwarden Send (I'm on a paid acct), depending on the person.

Update: I see that there are many ways to do this accountless. For some reason the search result was iroh.computer and I made some assumptions. There are definitely some iroh-based options that look like they would have essentially the same UX that I'm going for. Would love to hear if there's some specific pros or cons to croc or iroh.

Congrats on the launch! I'm glad more progress is being made in this area.

Because of the hallucinations inherent in transformer models, I went looking for a transducer-based model with a low WER. I have been super pleased with parakeet-unified-en-0.6b. It's WER isn't as low as Canto, but it's about as low as you can get (~5-6.5%) with a non-transformer-based model as far as I'm aware.

I've been very pleased with its output.

I wasn't looking for this, but it's also lightweight enough to run on my little potato PC, which has an i5 8th gen processor, and still transcribe 9-10x faster than realtime.

I vibe-coded a little wrapper for it and use it on folders of audio or podcast rss feeds or even YT playlists and channels and it's been one of my new favorite tools.

If you like Snapdrop, you'll love croc!

I used Snapdrop before they got bought and found it pretty reliable. It sometimes would not transfer and I went exploring.

I used magic-wormhole-based Warp and liked it, but now I use croc of getcroc.com for almost everything. Whether it's sending files to myself or others, it's just the most flexible and seems to be reliable and fast from my use.

(I haven't been using it super long, but I heartily recommend.)

What's under the hood:

"These new capabilities are powered by the next generation of Apple Foundation Models, custom-built in collaboration with Google and its Gemini models for deeply integrated Apple Intelligence experiences. These latest models run on device and on servers using Private Cloud Compute."

TIL about <meta http-equiv="X-Clacks-Overhead" content="GNU Terry Pratchett"> which is delightful and an example of what I love best about the Internet.

Thanks! I wasn't familiar with it (using FUTO now), but it looks really good.

Such as?

I love that little guy! lol

IANAL, but it might be relevant to others trying to answer what jurisdiction you are in. (US, EU, CN, JP, elsewhere?)

Also, when you bought from PianoDisc, did you agree to abide by a certain jurisdiction's laws in your use of PianoDisc? And did you explicitly agree to not share any sort of decoder/encoder in any ToS?

If you would like a 3rd party x.com web viewer to survive, probably best to not call attention to it atm.

Yes! And do you get the impression, like I do, that model effort and pelican effort seem correlated?

No.

Tautologies are beautiful and helpful because they help people and are full of beauty.

Except for the cases where no writing is better than LLM writing. The author's take seems ok to me, but what I struggle with most in AI writing is that it removes or abstracts away the intent of the original author. Not in every sentence and in every thought. But when a human says something, I can reason about why they said it the way they did. There are threads worth tugging on. 2 easy-to-understand examples: 1. when an ESL writer communicates something in awkward English, it's a fascinating glimpse into how they normally communicate about concepts. The meaning is usually not too difficult to infer and I can see how their language or culture affects how they see mine. It's very fun. 2. if an expert explains something and I get lost, that's usually a hint that my mental model could change to see the problem/solution more like they do.

But when a person expert or no uses an llm, information is being lost. If I misunderstand an llm's sentence, it is possible that "it knows what it's talking about", but with how frequently they don't know what they're talking about, I usually don't try and move on. It's not worth it.

In America, at least, though perhaps this applies more broadly, so many trends over the past decades all converge towards the lowering of the baseline levels of trust you should give towards a stranger (whether valid or not!). And llms have been a net-negative so far. We have exchanged some amount of trust for intelligence. And it didn't have to be an exchange.

I care less about llm disclosure and more about this: If you use an LLM for writing, please think sincerely about whether the output accurately represents what you would say and how you would say it to another person. Please avoid serving others nothingburgers and wordsalads. This has always been possible and it's easy to do it without realizing it with LLMs. The output can survive a scan, but dissolves upon further inspection.

tldr; make sure that what you publish is trustworthy.

Similarly, May gives me "8" when I think she gave pokeballs?

Slop disguised as satire. No, it was not fun.

It's funny how the life transformation in this article is: quitting a job and breaking up with a kind boyfriend against the advice of a parent. These aren't wrong, but... I'm not sold this is an improvement. Maybe this article is targeted at people who already know the author and trust her?

Looking at her bio, the only family connection I can find is, "Proud Mommy". For a "happiness and mortality expert", I would've expected her to say more in her bios across her sites about her loving husband and children. I'm not saying she's bad, I'm just saying that it would be more helpful if she, in general and in this article, talked more about what she's done outside of being a happiness expert. Why should we trust you? How do we know breaking up with this boyfriend was good? Even a comment about how that freed her up to find someone else who was truly better and worth committing to would be better than just selling us on this breakup on basis of her feeling they weren't right.

(And to be clear, I think feelings here are important. You don't have to enter into a relationship if you're not feeling it. A subconscious feeling can be onto something long before your mind is. But I feel that once you're in a relationship that's that intimate, feelings alone shouldn't end it.)

Always has been.

I am glad someone is doing something like this. Very cool to see a free market solution. I do wonder, how will they handle Instagram, Youtube, and X.com, the everything (incl. porn!) app?

Will definitely discontinue my Pro subscription if it's true. Claude Code is the only reason I signed up in the first place.

Helpful poison?

>> a quote from the article

> An opinionated comment or personal anecdote.

Counter opinion or added nuance. [1]

[1] A link for support or to demonstrate a counterexample.

I'm not getting enough sleep. I read this as ViltrumiteOS. I haven't even watched Invincibles.

I love yazi! My favorite file manager GUI or TUI. Very fast and scriptable.

Agreed.

It is also often a very bold claim to say "There is *no* evidence for X." The cases where you can truly say that are rare compared to how easily people (esp. non-experts) will say it. More often, it's more accurate to say, "I don't believe the arguments for X or evidence of X."

Respectfully,

Claude 3 Opus: $15.00 (Input) / $75.00 (Output) per 1M tokens

Claude 4 Opus: $15.00 (Input) / $75.00 (Output) per 1M tokens

Claude 4.1 Opus: $15.00 (Input) / $75.00 (Output) per 1M tokens

Claude 4.5 Opus: $5.00 (Input) / $25.00 (Output) per 1M tokens

This is thoughtfully done! I use canonreader.com on the web and The Literal Word bible app as an actual app that has one of the best UIs for reading. AndBible is really good for breadth and quality of resources.

I will add this to my rotation.

Also, I love to see a variety of fonts and the WEB version here.

Thanks for making this!

I can see the image just fine on Thorium!

Claude Code has been around for just under a year.

10k/365 = 27.4hrs per day. Respectfully, I don't think the author has had almost 10k hours with Claude Code.