it's wild that we saw programming languages go from syntactical to plain english in 2 years I needed a niche slack plugin this morning - spent 30 mins and I have it now It takes a thread and exports it to markdown so you can feed it to an agent elsewhere - useful when you talk about something on slack, have an agent running and want to give it more context while it's already working on something ps - this only works for "threads" right now https://github.com/rishavmukherji/slack-thread-export

1 reply

3 recasts

87 reactions

just set up clawdbot on an old machine and it's working well personal assistant tasks usually require little compute so didn't really feel compelled to do the new Mac Mini setup that everyone on AI Twitter is pitching using an Anthropic API key, you can get quite far on old hardware

26 replies

5 recasts

118 reactions

Fun fact got interviewed by the agent that we built ourselves (aka @neynar) thanks to @gmfarcaster might be the first of its kind

8 replies

10 recasts

104 reactions

most people should be looking at prediction markets for news, not necessarily betting on them I read somewhere that @barmstrong said prediction markets benefit from insiders betting (not sure if he actually said that or was made up). If he did, I think he is right. - prediction markets aren’t securities; they’re event contracts where payouts depend on outcomes, not managerial effort - insiders betting on prediction markets exposes truth faster to the world than waiting on traditional announcements - information has a price, prediction markets allow insiders to monetize their information (whether you like it or not) unless you have insider information to know the outcome of the market, betting on prediction markets is the same as gambling. there's nothing wrong with it, gambling can be fun (people play poker with friends) but it's good to be honest about it. Without added info, there's no added edge. However, as truth seeking machines, it's great to observe them and know the future. They are essentially, and obviously, time machines. It's funny in the sense that if someone had decided to build a time machine 20 years ago, they would not have predicted gambling as a tradeoff on the other end. one can argue "they normalize gambling" etc. however I don't think there's any reason to blame prediction markets solely for it. they are just the type of new market that went mainstream recently. humans have consistently increased the number of markets they have access to and have gambled on them. Imagine not having equity or commodity markets. Someone in the 1500s might have resisted against them for the same reasons. Kids born today will not find prediction markets any weirder than we find equity markets. We still have to teach kids about the risk of gambling ofc. It's possible to gamble on equities and it's possible to gamble on predictions. The options available have increased but the fundamentals are the same. Time machines are cool. Worth being picky about which markets you bet on.

34 replies

49 recasts

436 reactions

still wild that we can build full software products without ever opening an IDE imagine ten years from now

5 replies

4 recasts

80 reactions

reducing wealth gap is a broken concept do we look at star systems and be like all stars should be within a certain size of one another? income equality as a concept is lost, inequality will only increase. there is no closing the gap. there will be people so much wealthier than you that it'll be incomprehensible. they won't own mansions, they will own planets, galactic scale neighborhoods. just as we spend no time thinking about how other earth like planets in cosmos might be bigger or better, we will spend no time thinking about such absurd wealth. inequality cannot be stopped, it can only go up. only thing that matters now is increasing prosperity for everyone and is a goal worth putting effort behind. happy new year!

17 replies

4 recasts

59 reactions

I don't think content coins should be about "trading" the way humans trade today If content coins are price finding mechanisms for content, what they measure is - when was attention focused on that piece of content (x axis aka time) - How much attention did it receive (y axis aka price) Together the graph has the attention information needed for any piece of content I'm not sure humans will do the majority of this price finding manually. Most clearly don't like it. I think if it works, it'll have to be automated. One possible approach * - clients buy $0.000x for every second when the content is in the viewport** - They sell when it leaves the viewport - the buying and selling can be between users if users have small pre allocations set that the client can use - Users will sell to other users automatically with the above mechanism - Those early will make a few cents passively - The trading graph will show the attention data - Any real trader can still choose to manually trade and take on a bigger position Today content coins are trading engagement but they should really be looking at the viewport and trading true attention. Alongside automation, it might also need to be abstracted away. YouTube doesn't show creator earnings from ads in real time to the viewer. Crypto allows showing this real time but there's really no need for it to be front and center. It can be a layer deeper so only those interested dive in. This "price finding" work will make some money for the creators, some money for the platform. For human scrollers, some will make a few cents and some will lose a few cents. However, given average scrolling behavior - any single person scrolling through the feed should net out break even by being early on some and late on the others. No change to non trader behavior (and they notice nothing because the trading is automated behind the scenes). * obviously a thought experiment and I'm sure there's a bunch of things wrong with this approach that need to be refined ** Yes, some will automate scrolling to try and game this but that already happens today for generating fake view numbers so nothing new there ps - if anyone read this far, feel free to disagree with all of it. I personally have no vested interest in any specific type of coin, just had some time to think about this today

14 replies

8 recasts

60 reactions

there is this constant debate of whether everything will be onchain in the year 1995 > the tech-forward might have predicted most experiences to be online with AR/VR; today we're nowhere close and IRL experiences are growing > most, tech-forward or slow-to-catch-up-mainstream, would not have predicted tiktok, whoop, kalshi same thing applies to blockchains, most likely outcome is that - if you think onchain will be limited to a specific set, you're wrong - if you think everything will be onchain, you're wrong in fact, only thing we know with high probability is that our predictions (either for or against) for the next two decades will be mostly wrong

8 replies

0 recast

34 reactions

so AI founders who do well will have to - sell (5+n)% of privately held equity over 5 years* - *where n accounts for tax on the sale itself - find private buyers for a non public company - convince their board to allow that sale and deal with the cap table changes of bringing on the new buyers ?!?! I imagine this tax is to close the loophole where the ultra wealthy borrow against their equity without ever paying taxes? If the state absolutely has to enforce a tax, why not tax the borrowed amount instead of taxing unrealized gains? what am I missing? https://x.com/RoKhanna/status/2004776831501947267

7 replies

1 recast

16 reactions

AI conversations > podcasts for long drives Spent an hour talking to ChatGPT two nights ago. Stayed alert, asked follow-ups, no drowsiness. Talked ancient history, philosophy, random tangents. Felt completely natural. Wild that this seemed impossible 3 years ago and now it's just... normal? LLMs are still wildly underrated.

11 replies

12 recasts

100 reactions

complaints about - fb newsfeed launch were on fb - twitter content moderation were on twitter idt controversy has ever killed a social network

4 replies

0 recast

13 reactions

Martin Parr was a friend of my photography professor Chris Killip. Worked closely with Chris for many years and studied a lot of Martin’s work. Both Chris and Martin have passed now which makes me quite sad. They left a lasting impression on how I view the world and the sense of humor I grew to appreciate and embody. > “I realized pretty early on that controversy didn’t do you any harm,” Parr said. “And now, of course, it’s seemed to have all settled, and people are much more appreciative of that body of work” https://apple.news/AOCU51h1zQPa3orLadoFXqg

2 replies

3 recasts

36 reactions

every once in a while I am still blown away by how fast software engineering can move with LLMs imagine it being taken away, the world would be instantly worse off the comparisons to electricity or transistors is so apt, we can't imagine living without it

8 replies

9 recasts

72 reactions

granola knows more about me than I expected

7 replies

2 recasts

59 reactions

scared to disclose my Spotify wrapped age

12 replies

6 recasts

52 reactions