What have you been doing lately for your own fitness? Try out any new programs or exercises? Have any questions for others about your training? Want to vent about poor behavior in the gym? Started a new diet or have a new recipe you want to share? Anything else health and wellness related?

What have you been watching and reading this week? You don't need to give us a whole essay if you don't want to, but please write something! Feel free to talk about something you saw that was cool, something that was bad, ask for recommendations, or anything else you can think of.

If you want to, feel free to find the thing you're talking about and link to its pages on Anilist, MAL, or any other database you use!

In all the hype and debate and excitement and frustration around AI, I think sometimes the fact that we've achieved something really cool gets lost. I want to talk about that a bit, which I'll try to do in a simple way, without too much math or computer science.

There's a surprising amount of intelligence encoded in applied language. To me that's the most remarkable insight in LLM technology. Except it's not really surprising at all. The whole point of language is to encode intelligence. Communication, at its core, is about sharing intelligence. LLMs agents are a way of turning the intelligence encoded in language, particularly written language, into useful work.

You might imagine that language is just the starting point when building LLMs. That it gets converted into computer stuff and then it's not language anymore. But in a way it's language all the way down. Distill language into math and then run gradient descent on it and the intelligence (for lack of a better term) bubbles out.

Forgive that hyperbole, but there is something a little bit magical about it, and the technology is entirely unmagical. The magic is in the language.

At the core the tech is pretty simple. First you go and get all of the language. Just completely idiotic amounts of applied language. Because there's a threshold: if you don't use enough language you don't get a useful result. The more language you use, assuming it's good quality, the more useful the result.

After you have all of the language, you convert it into tokens, which are essentially pieces of words represented as numbers. You also compute a compressed version of the relationships between the tokens. The frontier labs don't say how much this costs publicly anymore but it's safe to say hundreds of millions in compute are required for this process. Each time you do it.

Once you've determined the token relationships, you've pre trained an LLM. You have the initial weights. At this point your LLM can output credible prose pretty reliably when you run inference on the weights. Meaning that you can take a given set of language tokens (a prompt) and determine which token should come next, over and over again, until you have a response. The fluency of the prose can be uncanny, and there are latent tendencies towards useful output, but you can't really do much with it yet.

The next step, fine tuning, modifies the weights in order to nudge the inference towards useful things like instruction following, tool use and reasoning. Or a simulacrum of reasoning. Fine tuning is a massive, multi step, iterative process that turns your LLM into a useful tool that can do more than output believable prose.

Something that's really fascinating to me about fine tuning is that what you're training is still the language relationship weights, and what comes out during inference is still language. But now it's language that is more likely to result in useful output and behaviors. Language is the substrate that LLMs reason through. A model that's more likely to infer the words "but what if I look at it another way" is more likely to consider multiple possibilities. A model that reliably infers language around tool use in appropriate situations is an agent. The behavior is encoded in the language.

After tens of millions in fine tuning costs, assuming you got it right, you have the core of a LLM that can function as an AI agent. You can put it in a harness and ask it to do things in the language of your choice and it will actually be able to do them a shockingly high percentage of the time. It can even do things that weren't explicitly part of the training. That's something which has never happened before outside of science fiction. Until recently it was one of those computer sciencey things that was probably going to happen someday, after we were all dead.

It sucks that this tech revolution is being driven by the capital class, and that it's happening as a mad dash for market domination and golden IPOs. In a more perfect world technology derived from collective human intelligence would be a public good rather than a profit driver.

But nevertheless, the ability to imprint some part of human intelligence into an autonomous agent is a remarkable achievement. One that we can't yet see the full size of. And it's possible not because we typed instructions into an interface, but because we've spent centuries encoding our intelligence into language.

Have you watched any TV shows recently you want to discuss? Any shows you want to recommend or are hyped about? Feel free to discuss anything here.

Please just try to provide fair warning of spoilers if you can.

This topic is for the Three Cheers for Tildes mobile app.

I'll summarize the major updates at the start of each similar topic, so people can read the updates and then hit Ignore if they don't care about more frequent updates and user feedback.

What's new:

Version 1.7.0 (Jul 27, 2026):

• Detect expired sessions and prompt to re-login
• [iOS] Combined external browser settings for clarity

[iOS] Version 1.7.1 (Jul 31, 2026): Added light/dark theme setting

On iOS, currently TestFlight only for the next week or so.

Version 1.7 fixes a long-standing problem in the app: the interface pretty much completely broke when login sessions expired (after a year), with no obvious way to fix it from the user's perspective. Now the app will detect expired sessions and show a prompt to re-login.

I promoted this from quiet bugfix to a more prominent release, because it was pretty terrible user experience, and anyone using the app for more than a year has run into it.

I had to test thoroughly on Android and iOS, since I don't want to ever break users' legitimate login sessions. I also had to change approaches partway: originally I wanted the app to automatically re-login, but that doesn't work for 2FA-enabled users. Also there were too many error-prone edge cases when trying automatic re-login, whereas a user-visible re-login prompt avoids most of those silent error cases.

Previously:

[Android] Version 1.6.8 (Jul 15, 2026): Fixed crash when searching for topics on Android 12 and earlier.

[Android] Version 1.6.7 (Jul 9, 2026): Fixed comment interactions after double-tap to vote.

[Android] Version 1.6.4 (Jun 18, 2026): Indicate when replying to a deleted comment, fixed minor UI bugs in topics feed, fixed networking bugs, fixed rare crashes, support Android 17

[Android] Version 1.6.3 (May 20, 2026): Added "Find in comments" to "..." menu. Fixed first search position in comments. Fixed tapping links in collapsed details summary. Fixed stale vote rendering.

[iOS] Version 1.6.3 (Jun 27, 2026): Indicate when replying to a deleted comment.

[iOS] Version 1.6.2 (Jun 14, 2026): Fixed comment scroll bugs, fixed networking bugs

[iOS] Version 1.6.1 (May 21, 2026): Improves on the Find in Comments feature. Fixes some UI bugs related to the Find Comment bar, and with potentially stale votes showing in the UI. Also adds the iPad pane toggle on iPadOS 18 and earlier, to bring the behavior closer to iPadOS 26, and fixes some iPad animation bugs.

Previous topic: May 2026

Where to get it

Android version on Google Play Store: https://play.google.com/store/apps/details?id=com.talklittle.android.tildes

Or sideloadable APK at https://www.talklittle.com/three-cheers/

iOS version on the App Store: https://apps.apple.com/app/three-cheers-for-tildes/id6470950557

Join TestFlight for iOS beta testing: https://testflight.apple.com/join/mpVk1qIy

The fragrance survey thread inspired me to take a perfume journey this weekend. I sniffed myself dizzy and discovered a dozen candidate scents and one odd piece of website copy.

There is something almost admirable about the rank fiction that serves as Eight & Bob’s origin story.

The gentleman enthusiast who winces at the thought of profit tainting his art. A recipe lost for half a century, recovered. A source plant so reclusive it escapes scientific classification; so temperamental, yet so indispensable, that only seven percent of its already fitful, dithering harvest dictates the global supply of a perfume from which the producer apportions each nation’s allowance in a grand concession to the yearning noses of the world….

I am a little inured to the extrait de bull in branding, but this packed potpourri of horseshit projects like the drydown of a cabbage dinner. And it just keeps delivering. There’s JFK incognito! An ever-resourceful butler named Philippe! A car crash. Nazis.

The point, I think, is to congratulate the buyer discerning enough to look past brand stories. All such tales are a volley of gas for the connoisseur to ignore, they seem to be saying. It might have made for a fun comic book. As it is, Eight & Bob’s puffery is a little too on the nose.

My real-world impression: I would like my linen cabinet to smell like this—faintly, maybe. There is an initial citrus fizz over cream soda in a clean white shirt; a soft-palmed, dry handshake that keeps a steady grip longer than you would expect; and cool wood. I'm going back to try it again. This is a scent that convinces you, "You don't need to check my references." And maybe Mother Teresa can show up in the backstory sequel.

Asked in ~comp thread, figured it should be its own post.

Thinking about whether I want to do some upgrades to my 10 year old PC or buy new. I could bump RAM to 16 without breaking the bank since its DDR4, and maybe splurge on a new processor, but I'm not sure if I'd just want to go with a new machine at that point, thoughts?

AMD Ryzen 3 1300
8.00GB Dual-Channel DDR4
MSI B350M PRO-VD PLUS
4GB ATI Radeon RX 480

Edit: To follow up some questions about what I'm using this computer for, not a whole lot at the moment. I was thinking of making it into a media server or a Steam OS console now that it's available to download, but it's more of a whim and a project itch to keep my mind off of personal and global events.

I live in an apartment on the third floor, and have a cat. She loves watching birds from our balcony, so we got her a bird-feeder to put in our balcony.

It has done a great job of bringing in more birds (primarily finches and sparrows) for her to watch, except the sunflower seeds we got for it result in an extreme mess all over the balcony and I worry our downstairs neighbors likely hate us as a result, so we’ve taken it down.

Wondering if anyone has recommendations for a bird feeder or bird seed that would be less messy? Or should this just turn into a “when we move to a first-floor apartment or a house” sorta dream.

Unfortunately, my wife is allergic to peanuts, so any bird seed that is made with peanuts (and from what we’ve seen, almost all are) is not an option.

Came to post and saw this thread popped up 5 hours prior (okay, near 7 hours as I'm still trying to look). So please help them first!

Sadly, while a lot of the requirements are the same, it doesn't offer what I'm looking for. Here are the demands (and here is why I cannot profit from it):

• USB A or C wired (no BT or detachable connection)
• Full 100% size

Similar demands are:

• Works with Linux
• Cherry MX (or comparable) switches - red or brown level
• Backlit with custom coloring

Honestly, Ducky seemed to be exactly what I was looking for, but I see no "buy" option, nor are there Linux options. Yeah, they're on mechanicalkeyboards.com but the third party is disconcerting and also the options I want are out of stock.

This is a recurring post to discuss programming or other technical projects that we've been working on. Tell us about one of your recent projects, either at work or personal projects. What's interesting about it? Are you having trouble with anything?

Have you watched any movies recently you want to discuss? Any films you want to recommend or are hyped about? Feel free to discuss anything here.

Please just try to provide fair warning of spoilers if you can.