Like what you see? Subscribe here and get it every week in your inbox!
Issue #310 - February 16, 2025
If you are looking for work, check out this month's Who is hiring? and Who wants to be hired? threads.
Here are the top threads of the week, happy reading!
1. Ask HN: Former employees' RSUs at risk after startup's IPO
Top comment by vessenes
I read down pretty far and did not see this basic advice: you need a lawyer. Hire a very good one: they are cheaper than poor lawyers by an order of magnitude.
That lawyer will review your agreements, the state laws and the communications with the company and tell you where you’re at. You could all go in together for the lawyer btw if you have the same contract.
I bet that lawyer will tell you (if this is in California) that you just need to send a letter saying “cool guys, send the shares, I’ll worry about the taxes. I don’t consent to forfeiture.” But, crucially, I am not a lawyer and I haven’t read your agreements.
anyway I would not stress over this, but I would act quickly, don’t sign anything, don’t communicate with the company unt you’ve talked to the lawyer, and know that worst case you will be able to find a loan to bridge this. It may actually be a good candidate for a venture debt firm, but either way the lawyer you hire will probably have some leads.
Congrats on the IPO!
2. Ask HN: Is anybody building an alternative transformer?
Top comment by czhu12
The MAMBA [1] model gained some traction as a potential successor. It's basically an RNN without the non linearity applied across hidden states, which makes it logarithmic time (instead of linear time) inference with a parallelizable scan [2].
It promises much faster inference with much lower compute costs, and I think up to 7B params, performs on par with transformers. I've yet to see a 40B+ model trained.
The researches of MAMBA went on to start a company called Cartesia [3], which is MAMBA applied to voice models
[1] https://jackcook.com/2024/02/23/mamba.html
[2] https://www.csd.uwo.ca/~mmorenom/HPC-Slides/Parallel_prefix_... <- Pulled up a random example from google, but Stanford CS149 has an entire lecture devoted to parallel scan.
[3] https://cartesia.ai/
3. Ask HN: What's the best implementation of Conway's Game of Life?
Top comment by selcuka
This one is quite interesting:
https://oimo.io/works/life/
It was also featured on HN:
https://news.ycombinator.com/item?id=39799755
4. Ask HN: Physics PhD at Stanford or Berkeley
Top comment by GlibMonkeyDeath
Another Ph.D. physicist here (Ivy, not Stanford/Berkeley)
As others have pointed out, your prospective advisor(s) are the most important thing to consider. You can't go wrong with either school.
That said, when choosing an advisor:
* Pay attention to where the advisor's former students ended up. The former students are a natural "network" for you when you graduate. If you can, ask relatively recent grads about their experience.
* Meet the prospective advisor's current students and post-docs - are they happy? Will you fit in with them? Do they graduate in a reasonable amount of time? Ask other grad students about the professor as well. Trust me, each professor is going to have a reputation.
* If you want to stay in academia, mid-career advisors are the "safest" - an assistant prof may be working on something exciting, but the research will probably be more risky, and the professor might even have to leave mid-way through your thesis work if they don't get tenure. A late-career advisor may presently sit on a lot of committees and be more well-known, but by the time you need their recommendation for jobs/tenure they may have considerably less influence (that happened to me, although it was fine in the end.)
Read Feibelman's "A PhD is not enough" - still lots of good advice even though written 30+ years ago.
5. Ask HN: What country would you like to relocate to and why?
Top comment by bradhe
I’m an American living in Germany for the past 4 years. I moved here from the west coast after I sold my last startup because I wanted a new, different kind of adventure. I don’t see myself going back to the US any time soon, but I’m actively working on relocating to the UK.
My wife and I want to move there because we want to be somewhere more English speaking. I speak German well enough, but she’s really struggled. Truth is, it’s really difficult to integrate into a society if you don’t speak the language at a high level. And after around year 2.5 or 3 I really started thinking hard about integration.
The political situation in Germany is a bit of a struggle, but coming from the US…this is a cake walk. I’m so happy to not have to deal with American politics on the regular.
Economically speaking, there are obviously way more opportunities in the UK, especially for my wife who works in HR given the language requirement. In the tech ecosystem in Germany, it’s two-tiered: The really good engineers make a wage comparable to the US. Think like tier 2 or tier 3 cities like Dallas or Miami. I says it’s about 70% of the Bay Area. But the VAST majority of engineers, working for German companies, make way less. Average wage is probably €70k?
Obviously, your euro goes massively further here as compared to the UK and US. My point in all this is that it’s really difficult to actually compare apples to apples, what your quality of life will be based on your salary and cost of living alone in different parts of the world.
6. Ask HN: What intelligent forums exist outside of HN?
Top comment by spiderfarmer
I know you're probably looking for English language websites, but if you just want the satisfaction of knowing that forums are still "a thing":
This is a Dutch forum for Dairy Farmers. It's at the root of a lot of innovations:
https://www.prikkebord.nl
This is a Dutch forum for tractor enthusiasts. The people there are very helpful and you can basically restore a tractor as a newbie with their help:
https://www.tractorfan.nl
Even though these seem like very specific niches, both websites have tens of thousands of daily visitors.
7. Ask HN: Finding interesting work if disenchanted with "big tech" & "VC-backed"
Top comment by andrewfromx
CS Degree from 1996, been working for startup then big company, startup then big company, over and over for, yikes 30 years. I'm now at big company again and I've completely lost the urge to "change the world" and the product my big company makes isn't even a product I use myself. (Not that it's bad, just not my thing.) I'm in it now for the people and the tech. You'll never find that fulfillment you are seeking in working for the right company doing what you think is important. Just pick one and fall in love with your co-workers, make them friends, and do tech for the sake of tech.
8. Ask HN: Are AI dev tools lowering the barrier to entry for creating software?
Top comment by sevensor
Every summer, my community pool has a cardboard regatta. Kids can use as much duct tape as they want to waterproof a cardboard box and paddle it 25 yards to the other side. Half of the vessels sink within a length or two and the kids have to swim to the edge of the pool. There’s no age limit, and last year a grown man entered a fully engineered catamaran design that beat all the others handily. The secret was using way more duct tape than anybody else.
AI dev tools are that catamaran. They’ll get you across the pool; you might even get half a mile from shore, but there you are, in the middle of the lake, sitting on cardboard and duct tape, wishing you knew how to swim.
Top comment by rozenmd
I felt the same way as a naive undergrad - "every piece of code worth writing is already in a library you can import". Very clearly not the case.
When it comes to SaaS, no one cares about AI vs non AI. What your customer cares about is "do you solve this problem that makes my job hell?"/"does this tool make me awesome at my job?"
How do I know this? I built the 200th uptime monitoring alternative to Pingdom four years ago, and I'm still getting customers in this age of AI.
10. Ask HN: Wantrepreneur who's run out of energy/ideas. What now?
Top comment by mrdependable
Build something that there is already a market for but focus on a specific niche of customers and build it only for them. There are a lot of businesses that struggle to get software to work the way they need it to. For example, inventory tracking for flower shops.