CEO @rocicorp. Building @zero__ms, raising two great kids, trying to be a better person. Past: @replicache, @nomsdb, @googlechrome, Greasemonkey, misc OSS.

Hawaii, USA
Sync engines enable instantly responsive UI. But the tradeoff is usually slow startup because all data must be pre-loaded. Zero's query-driven sync enables precise control over when data syncs. To demonstrate, we scaled our "zbugs" demo up to 2.5M rows. Introducing Gigabugs:
9
16
414
57,743
Cloudflare’s platform naming remains undefeated: DO and D1 are both things, but are completely unrelated components. Not different versions of “D”. Amazing. Legendary. Nobody names like Cloudflare.
12
1
101
14,417
(with thanks and apologies to @TTomTToro)
314
I'm telling ya'll - you're going to wish you'd been nice to them.
10
832
Quiz: Is IVM more efficient as write rate goes up or down? Why?
3
7
2,320
We think of IVM as being good for write-heavy systems but that is actually exactly opposite. IVM is good for read-heavy systems! What we're really thinking of when we intuitively think of IVM being good for writes is *freshness*. When we want to know ASAP about a change, IVM is good. Because otherwise, we'd have to poll, driving read rate through the roof.
1
2
291
It's the reads that matter, not the writes.
1
248
Andreessen used to say that software was eating the world. He said it because it was already clear that software was going to eventually automate all of human civilization. That was when humans were writing the software. Even at that plodding, expensive, human pace, there was already a clear demand for software to take over everything. So it's obvious in retrospect that software was the killer app for AI. There was already infinite demand. The same incentives that drove software into every corner of our world make software the obvious path for AI. It's just the rate that has changed. Software is no longer eating the world - it's inhaling it.
3
19
2,038
Here is some trivia for you. One of the main people behind this also started Chrome and Chrome OS. And was my boss for ~10 years. Can you imagine building the world's most popular browser _and_ destroying mosquitos in a single life? King shit.
When I stilled worked on biosecurity, I first learned about Debug, a Google project to reduce mosquito populations. Results from their Singapore pilot are now out. They crushed mosquito populations, reducing dengue infections by 70% in the process:
4
7
211
12,182
Did not mean to imply that being my boss ranks up there with building software used by billions of people or saving millions of lives. But otoh…
2
433
Every working programmer should get a Tesla with FSD if you can. I think it is an excellent way to engage with AI in a different context and get your mind thinking about what work will become. Currently (probably not for very long) but currently, FSD and I "pair drive". FSD is better at a lot of things, I'm better at some things. Often we're best together, but also often I just let it handle the drudge work.
25
2,178
Everyone should read _Red Mars_ if you haven't already. Incredibly beautiful stories, and more applicable now than ever. PS - when I look at these, immediate claustrophobia. Something deep inside me needs access to outside. Looking at outside not enough, I need to actually be able to go out there. And I don't think EVA would do it for me. I need to breathe the air.
Cosmic ray coverage for Mars architecture unlocked 🫡
9
2,222
I started programming opportunistically. I went to school for music and had a ska/punk band, but that didn't pay very well. The web happened and I was comfortable on computers so taught myself webdev. I didn't consider myself a programmer, though. I was a skater, a musician. This was just a job. I'd ride it as long as it lasted, then do something else. "I'm a smart guy", I'd tell myself. "I can learn new things". I had a financially insecure childhood, and as a result could never really relax, even as I was earning more and more money. I always expected the bottom to drop out. When the dot-com crash came around, I wasn't surprised at all -- it was what I'd expected. I moved on. Microsoft was pushing .NET really hard so I learned that to get my next job. I didn't consider myself a .NET programmer though, or any kind of programmer. Just a job. Bottom can drop out anytime. I'll get as much money as I can then learn the next thing. Then web 2.0 came around and .NET started to sour. So I moved back to the web. I was never precious about it. Over time, I did actually grow to love the work. And got more and more of my identity wrapped up in it. But at the core, always, was "this is just a job, it can end any time, I'm a smart guy, I can learn new things". By the time 2008 came around, it was a little scarier. I'd built more of my life around programming. Committed harder. But still, because of that deep financial anxiety from childhood, I'd saved almost everything. I was prepared. The markets crashed, but programming didn't go anywhere. It kept growing. Over decades I came to trust that. I came to finally admit to myself: I am a programmer. This is what I love. I will do this until I die. When my kids were born, I assumed they would be programmers too. My son wrote his first JS in third or fourth grade. He loved it. I figured he'd be safe. Software was eating the world. I grew complacent. How ironic that here now, after 3 decades, the thing I worried about all those years ago finally arrived. Programming, the way we've done it for almost a century, is dying. The old defense mechanism is still there though. A little dusty, but I can brush it off. This was just a job. I'm a smart guy. I can learn new things. I am not (just) a programmer.
17
2
160
16,669
There is going to be a bit of a weird hangover where we have the last remaining software developers strongly passionate about languages and frameworks when nobody even touches them hands-on anymore. Like being passionate about which supplier your car manufacturer uses for wiring.
17
4
106
26,878
Yup. Didn't really expect this 2y ago but I predict Rust takes over ~all of server programming and lots of client programming too. We are just beginning to figure out the massive impacts from agentic programming. But sort of obviously, DX doesn't matter in the same way so a lot of the old assumptions are dead. JS still needed in some cases structurally -- SSR is the obvious one. JS needed on server for pages because it's needed on client. For now. JS also more amendable to runtime extension (eg plugins). But the lid really is blown off. Hard to predict where things go next more generally. Rust has a lot of performance/security/reliability benefits which manifest ultimately as *user* benefits. At the same time, its DX disadvantages don't matter as much anymore. Whatever happens it's going to happen fast because individual creative hackers are so much more highly empowered. Maybe somebody comes up with a compelling way to write pages with Rust? Maybe somebody improves TypeScript dramatically? Maybe hybrids become popular for runtime extension or people just use Rust somehow for that? Maybe some other language entirely? Who the fuck knows.
JavaScript was never designed for the server. If that doesn't change soon, JavaScript on the server will be replaced by Rust. The fix needs to happen in the engine. Sound types. Ahead-of-time compilation. Threads with shared objects.
47
7
160
19,678