Zulip is a critical piece of software for my business: It's the main tool my company has used to communicate for more than 13 years, and it's the primary forum used by our 3,000+ person alumni community. Our Zulip realm has over 4M messages.
I share this because I hope it makes it clear that I have a vested interest in Zulip's future. And I'm happy about this news; I'm confident Zulip will continue to improve for many years.
Also, for those who don't know: Zulip was initially a for-profit startup, which was acquired by Dropbox in 2014. Tim then went to great lengths to get Dropbox to later open source it, and allow him to found a new company (the one that was today donated to the new nonprofit foundation) to continue work on Zulip. I can't think of any other cases where a founder has gone to such great lengths to do right by their users.
> I'm certain that he's not doing this because he wants more money.
There are many reasons to change job. The pay is always one of them (if you don't work for money, it's not called a job, is it?).
> join Anthropic because of its remarkable commitment to the responsible development of AI for the long-term benefit of humanity.
Obviously, it's better to believe that what Anthropic is doing is good for humanity when you decide to go working for them. But it is at the very least debatable.
I was a part of the Zulip project as a contributor and have contributed > 100 commits to `zulip/zulip` and also admire Tim a ton. But, leaving you're life's work to work on "long-term benifit of humanity" at Anthropic doesn't sound right to me. I am guessing Tim isn't going to work on safety research or interpretability side of Anthropic, that's not his expertise. Hence, leaving Zulip to help build anthropic a new software is meh. There are labs who actually care about people and aren't pretentious like Anthropic. Nevertheless, wish him, alya and rest of the team all the best; they are genuinely nice people. I don't know if I'll have interest in sticking to the project anymore though (+ I am not sure about other core member's status like Anders -- that will affect my decision too).
> There are many reasons to change job. The pay is always one of them (if you don't work for money, it's not called a job, is it?).
Not at the same scale as this, but I've seen friends deliberately choose to get paid less, perhaps much less money, because they wanted to do something. Video games for example, does not pay well, but it may be your passion. Banking pays very well, but it's hard to find any significant emotional involvement.
You can probably argue that's what I did, but it's complicated because I'm hard work. I can't stand debt but I also don't like the feeling of not knowing how to spend all the money. I can say that it's surprisingly hard to get people who are hiring you to accept that (a) the number you put in their mandatory "previous salary" box is correct and yet (b) yes you did understand that they have fixed pay scales and can't possibly match that.
Yeah why else would a person choose to join an AI company right before an IPO worth trillions, almost guaranteeing any employer there to capture a massive multi-generational wealth defining bag, what €ould ₿¢ th¢ ₹ea$on I wonder?
If he thinks working for Anthropic is a good "cause" to devote his time to then that is also very disappointing. That would make him either very delusional as to the effects of Anthropic's work or naive in what he can achieve as their employee.
I've been a happy Zulip user (and realm admin) for 13 years: it's one of my favorite pieces of software, and I use it daily. My understanding is these changes will be very good for Zulip's long-term stability and success.
(I'm a volunteer member of the new foundation's advisory board.)
At least so far, the impact of LLMs has been overall positive for RC. We have a big-tent community, which includes everyone from those who avoid LLMs at all costs to those who are all-in on them, and everyone in between. We have folks who come to our retreats with the explicit goal of not programming with LLMs[1] (usually because they've been using them extensively at their previous jobs) and others who use RC as an opportunity to learn how to program effectively with AI. There's also a lot of discussion here about how (and to what extent) to use LLMs effectively in the context of learning.
We wrote up our position on AI last July, and while a lot has changed since then, I think all our conclusions still hold up well[2]: "...whether you choose to embrace or avoid AI in your work at RC, you will need to build your own mental structures to grow as a programmer. When using AI, use it to amplify your ambitions, not to abdicate your agency. And regardless of what you do, be curious about and kind to the people around you."
Like everyone else, I have no idea what the future holds, but I'm confident we'll find a way to navigate it. I'm sure I have some motivated reasoning here, but I really do believe that humans will still want to understand and build things no matter how good the models get.
Most folks do RC between jobs, either because they quit their job specifically to do RC or because they lost their job and then decide to apply. Other common ways are as part of a formal sabbatical (returning either to an industry job or to academia), as part of garden leave, or while on summer break (for college and grad students). We also get a fair number of freelancers/independent contractors (who stop doing their normal work during their batches), as well as some retirees.
Some folks use RC as a way to enter the industry (both new grads and folks switching careers), though the majority of people who attend have already worked professionally as programmers.
We've had people aged 12 to early 70s attend, though most Recursers are in their 20s, 30s, and 40s.
I've had the pleasure of following Peter's progress on this project over the past 18 months or so, and it's been incredible to see how far he's taken it.
When he first described it to me, I didn't really "get" it (is it a game? a simulation? some other sort of environment?), and it wasn't until seeing an actual demo and hearing Peter explain his thinking more deeply that it clicked.
It's basically a giant simulation environment that is 1) visually stunning (and all visual aspects are meaningful / carry semantic information and aren't just glitter), 2) technically quite impressive, and 3) built for rapid exploration and experimentation. If that sounds at all interesting, you should watch the video to hear Peter's talk!
(Writing this as someone who generally doesn't like to watch videos online; this is one of the rare cases where I think it's worth it, and a video is a better format than text in explaining the thing.)
What I really want, is to run and experiment with it myself, locally. But I couldn't find a repository anywhere, even from his linked GitHub profile. You happen to know if it's online somewhere?
Aw, shucks. Oh well, guess he at least have another follower, looks like really interesting stuff :) Looks like fun to play around with and get inspired by.
We were among the first Paper users (starting from the private beta). We loved the product for the first few years, but then it stagnated. We finally switched away from it a year and a half ago.
Their export feature has been broken for over a year. Support hasn't helped, and our data is still trapped.
I share this because I hope it makes it clear that I have a vested interest in Zulip's future. And I'm happy about this news; I'm confident Zulip will continue to improve for many years.
Also, for those who don't know: Zulip was initially a for-profit startup, which was acquired by Dropbox in 2014. Tim then went to great lengths to get Dropbox to later open source it, and allow him to found a new company (the one that was today donated to the new nonprofit foundation) to continue work on Zulip. I can't think of any other cases where a founder has gone to such great lengths to do right by their users.