Joe Fabisevich

Joe Fabisevich

@mergesort.me

Director of AI Product Engineering @ Pursuit.org. I used to make Twitter (better), then that all went to hell. Now I make puns, @plinky.app, and build.ms. Born and raised New Yorker, trying to do a little good and be the friend you made along the way. 🍕🐱⚾️

5.7K followers57 following6.7K posts

Pinned

Hi, I’m Joe! I used to work @ Twitter on all the hard things like misinformation, disinformation, harassment, and civic integrity. I also spent a ton of effort on little improvements to make people’s lives better. These days I’m an indie developer making it easy to save links for later, with Plinky.

plinky.appPlinkyPlinky helps you easily save links from any device, so you can enjoy them later. Save on your iPhone, iPad, Chrome, Safari, Firefox, Zapier, and so much more.

People seem surprised that folks like Gruber or Viticci didn't enjoy participating on Mastodon, but honestly I get it. There are lots of opinions you can't share on X, but if you actually believe AI is changing the way software is being built then Mastodon isn't going to be a great place for you.

Don't get me wrong, that isn't nearly enough incentive for me to post on X, but I can at least understand where they're coming from. X is a horribly unpleasant place to be, but Mastodon (and Bluesky for that matter) can be as well.

I don't say this lightly: this approach is clearly the future of software development. I have a blog post brewing that touches on the subject, but fuzz testing and formal verification at scale will make all software an order of magnitude better, safer, and more reliable.

anthropic.comLLMs are making formal software verification feasible at a much larger scaleClaude formally proved security properties in the compiled code of 123 open-source libraries. Along the way, it found bugs that had eluded detection for years.

Joe Fabisevich reposted

This is my first big update since I started my new job. Turns out having a job requires work, who knew? But I love Plinky, so I’ve continued working on it during my nights and weekends. This release is a labor of love, focusing on all the little things I’ve wanted to get to for quite some time.

Plinky 🔗@plinky.app

Plinky 6.1 is out today, and its headline feature is making everything a little better. There are over 25 fixes and improvements, ranging from performance to better consistency to longstanding bug fixes. But there are three particular improvements that stand out.

This is my first big update since I started my new job. Turns out having a job requires work, who knew? But I love Plinky, so I’ve continued working on it during my nights and weekends. This release is a labor of love, focusing on all the little things I’ve wanted to get to for quite some time.

Plinky 🔗@plinky.app

Plinky 6.1 is out today, and its headline feature is making everything a little better. There are over 25 fixes and improvements, ranging from performance to better consistency to longstanding bug fixes. But there are three particular improvements that stand out.

I’ve been wrestling with this in my head for a few days and can’t put a solid number to it, so I’m going to ask Bluesky instead. What number would you assign to the probability that in five years AI will be good enough that people will be paying to hire drop-in AI knowledge workers?

My number is 60-70%. It may not be the best worker you’ve ever had, but it will be good enough that small businesses will feel it’s worth the money. The implications of this on the economy are of course tremendous.

Pursuit already has an 85% success rate of helping people making an average of $19,000/year get jobs where they make at least $85,000/year. With this funding, we're going to be able to scale that to 40,000 New Yorkers by 2032 to do a whole lot of good for the people who need it most.

pursuit.orgWhy We're Building the AI Jobs Institute in QueensPursuit's AI Jobs Institute in Long Island City will serve up to 40,000 New Yorkers for AI careers by 2032. CEO Nick Simmons explains why we're building it.

Stratechery partnering with Stripe is fine, but the partnership also means there are now ad breaks in his podcasts and writing. That leaves a sour taste in my mouth. Ben Thompson says it's a positive because he can now make more content, but I've paid him for the quality not quantity of his content.

This is the indie version of enshittification. I'm not against people striving to do more, it's often admirable. But you shouldn't accept a notable drop in quality to make that happen. When you start doing that because the ends justify the means, you've actually lost the plot without realizing it.

I'm not that interested in an iPhone Duo, not because it's not a cool idea, but because as always I want a smaller phone. So what I'm saying is… John Ternus please gimme dat iPhone Duo Mini.

The Mastodon 5.0 design looks more Tumblr than Twitter, and I really dig it! Nice to see them take an approach to social media that is more differentiated than just "it's tweeting but federated".

1/ I’ve always thought type systems are amazing, and that we should move anything programs can express to the type system, to remove whole classes of errors. And as much as I hate on Swift Concurrency for being inscrutable, that’s always been the promise.

In retrospect, a world where agents can use the compiler to verify concurrency promises that most developers couldn’t internalize due to the inherent complexity is actually a pretty great outcome. I’m back to rooting hard for Swift Concurrency. 🥳

Me vibecoding: Amazing, the possibilities are endless. Other people vibecoding: Ew, look at that sloppity slop slop.

This may be the AI psychosis speaking but I've been at it with ChatGPT Pro for six hours and have discovered a better b-tree implementation. Luckily I know nothing about b-trees so my c-tree is guaranteed to be a completely novel implementation from first principles.

I hate that there still is no better mobile browser for iPhone than Arc Search — which was effectively abandoned two years ago.

Who has two thumbs and hasn't been able to fall asleep since 4am because his mind was racing all night brainstorming a new app? This very tired guy.

Who also had Codex slop together a proof of concept that validates the idea and deploy it over wifi so he didn't have to leave his cozy bed? This guy who's soon gonna regret last night.