Live data from Hacker News

Hard Startups

blog.samaltman.com

201–210 of 331 posts

Re: Hard Startups

#201

Earlier quoted context omitted.

Former fusion startup founder here. This is a very good point. My startup took 6 years to fail and much of our competition at the time is still at it 20 years later. There are a number of things which are also impedance mismatches between the typical SV startup mode and hard startups. One of these is the failure path. When an adtech startup fails, the founders just go get well-paying jobs at a FAANG, or at least that…

That came up recently in connection with a VR-related startup. They've been working for years, and they just canned the project and laid off about 60 people. It's all in-house code. So now the people laid off have to find new jobs. They don't have any of the buzzwords resume scanners want. Heroku? No. React? No. Unreal Engine? No. Unity? No. SpatialOS? No. WebAsm? No. Kubernetes? No. Docker? No. AWS? No. Javascript?…

> Maybe they can get a job fixing legacy C++ applications.

There's big money in that - those legacy applications are vital to the functioning of big corporations. Specialising in large in-house C++ applications obviously doesn't help develop a career in trendy webdev (and nor should it), but it absolutely gives you some good options.

Re: Hard Startups

#202
Probably the hardest startups today tacle climate related problems. They face hard problems and the problem of not yet marketed demands.

Re: Hard Startups

#203

Earlier quoted context omitted.

Former fusion startup founder here. This is a very good point. My startup took 6 years to fail and much of our competition at the time is still at it 20 years later. There are a number of things which are also impedance mismatches between the typical SV startup mode and hard startups. One of these is the failure path. When an adtech startup fails, the founders just go get well-paying jobs at a FAANG, or at least that…

That came up recently in connection with a VR-related startup. They've been working for years, and they just canned the project and laid off about 60 people. It's all in-house code. So now the people laid off have to find new jobs. They don't have any of the buzzwords resume scanners want. Heroku? No. React? No. Unreal Engine? No. Unity? No. SpatialOS? No. WebAsm? No. Kubernetes? No. Docker? No. AWS? No. Javascript?…

I used to work on an in-house framework. The trick is to write like "used proprietary JS framework xyz.js, a framework similar to react.js". The scanners will still pick it up.

Re: Hard Startups

#204

Earlier quoted context omitted.

There are a few examples of government agencies which can make many parallel bets, on the understanding that most will fail but a few may pay off big. But, by "a few examples", I actually just mean DARPA. I'm sure there are others, not many though. Politics does not work well with "most of these projects failed". It's too easy of a target. Anything that failed, is an easy target for a congressman to take aim at as be…

>But, the logic behind "many long-shot bets in parallel" is inherently more difficult to explain in the political arena, which is why I think VC may actually be better at some of these kinds of things. VC backed rockets have already taken longer to take a human to orbit than government money took to take a human to the moon. This is on top of the fact that they have had 60 years of technological improvement. Simply p…

The amount of money matters too. I'm sure if we were willing to plough 5% of the US budget into the VC approach we'd see bigger results.

Re: Hard Startups

#205

Earlier quoted context omitted.

That's a good question. My first thought was, "We didn't". But that's really just pessimistic memories speaking. My second thought was, "A perpetual state of stress and fear!". Also pessimistic memories, but closer to the truth. In reality, we had initial seed funding for about a year. We had a number of goals we had established, including vacuum chamber, getting the ion source running, first beam, first neutrons. We…

If you attempted another fusion startup today would the odds of success be higher, or is fusion still about as far away as it was 20 years ago?

There was a list posted of the companies around working on fusion recently:

https://www.fusionenergybase.com/organizations/

And an article about why this is the case from the same person:

https://www.fusionenergybase.com/article/the-number-of-fusio...

Re: Hard Startups

#206
post #16

Earlier quoted context omitted.

Right. Which is why government research is probably a better model for these world-changing basic technologies than Silicon Valley. Ugh, can you imagine how bad TCP/IP would have sucked if it had been developed by venture capital funding?

Is that true though? Nobody -- public, private, international -- has had any luck in fusion. I think that has more to do with the problem itself. But consider "world-changing" technologies like electric cars. The SV model doesn't see so terrible, does it? (Yes, I know about government grants. It's still a primarily private industry play.)

I saw a controversy recently in an open-source project where someone who packaged up the implementation with a friendly GUI was making thousands a month on Patreon, while none of that was going to the people who actually implemented the hard technical parts.

Credit for these technologies can seem similar - the government does all the basic research (through universities etc), funds the development of the tech through grants and the like, and Silicon Valley just does the customer-facing part and picks up all the credit.

Re: Hard Startups

#207
It would be nice if part of the ecosystem was structured more in terms of supporting founders who want to take on hard problems via way of dedicating their life to it. I'll provide my own perspective here in case anyone finds it interesting:

I've been working on my startup for 7 years now, and the damage it's done to my personal life and well-being has pretty much ground progress to a halt, with the exception of advances in thinking in certain areas, which has been nothing short of mind-blowing recently.

Of course, a large part of this is my own failure to plan instrumentally. Admittedly it's quite backwards to spend years planning the mid and high-levels first, then not have a solid plan to get from 0 to 1. Ending up there by accident is easier than you'd think.

Fortunately I now have a fully tractable, instrumental plan that goes from 0 to ripping the guts out of a couple FAANG-scale companies:

1. Get a well-paying engineering job I'll probably hate, because money, quality of life, and sanity all have instrumental value. Hopefully keep the fire alive during this period.

2. Work it for about two years, then raise a pre-seed round (at year 9!) to support myself for the next couple of years, while I embark on a formal pre-production phase, and possibly develop an up-to-date prototype/PoC.

3. Potentially a seed round totaling a few million. Hire a few key people and put together a refined prototype/pitch over the course of a year, for a Series A that no high-caliber VC with sufficient funds could refuse.

4. At year 12, raise a 100M+ Series A, with the caveat it's going to take about a decade to do properly, with further funding rounds along the way.

5. Complete the journey, or at least achieve the intended objectives, a decade later at year 22.

---

I'm well aware that sounds insane, mostly was just a writing exercise to underscore the timescales involved. The funding requirements are high because it's basically a blueprint for something that will become an indispensable part of people's every day life, and it consists of multiple verticals all woven together, sharing a common technology. Thinking about how to do this precisely was not trivial and took years.

End of the day I'd rather be skipping to step two, but I'm ironically out of time at this point. Job it is.

In retrospect, a proper plan would have been this:

a) Work a job until financially comfortable.

b) Do a smaller-scale startup or two.

c) Then dream big.

Hindsight is of course 20/20, and I suspect isolation, desperation, being depressed, being uncomfortable—all those things breed a level of ideation you might otherwise not encounter. It ain't recommended, though at least perhaps there's a silver lining.

---

For the most part, the tractability of the plan has only increased with time, not decreased, and I don't view competition or someone else figuring it out as a threat.

The biggest threat I face actually, is some of my talent pool may be aging out. Of the 50 or so people I've kept an eye on for years and basically dream of hiring on a daily basis, a few key people are in or approaching their 60s.

On the bright side, if anyone ever doubts how committed I am in the long-term, it should be easy enough to simply respond that I'm moving so slow the people I want to hire may end up retiring before I can hire them!

Re: Hard Startups

#208

Earlier quoted context omitted.

You envision a world with exactly one way to share photos? In this world, is there also only one word-sharing site? I think there are more information problems out there than software engineers.

Consider this: every day on ProductHunt, we see dozens if not hundred new apps pumped out. How many of those succeed? Not very many. Why? Each and every one of those apps was designed to solve a problem. So, why is there so little demand? It's because users already have another app/spreadsheet/website/solution/process that does something too similar and because their existing solution is good enough. The fact that so…

You're exactly right. Outside of deep deep tech (e.g. self-driving cars), you'd have to really, REALLY dig deep to find one or two startups that failed because of lack of technical talent.

Now consider how many startups failed because of lack of customer demand - nearly all of them.

This isn't just an indicator of an oversupply of talent, it also suggests an oversupply of capital. Cheap money lets you find talent, but it doesn't guarantee a path to a profitable business.

Re: Hard Startups

#209
Translation: the only startups willing to play YC’s sweatshop game are fruit flies; they want meatier deal flow. I doubt the founders of hard startups are that stupid.

Re: Hard Startups

#210
post #16

Earlier quoted context omitted.

Right. Which is why government research is probably a better model for these world-changing basic technologies than Silicon Valley. Ugh, can you imagine how bad TCP/IP would have sucked if it had been developed by venture capital funding?

There are a few examples of government agencies which can make many parallel bets, on the understanding that most will fail but a few may pay off big. But, by "a few examples", I actually just mean DARPA. I'm sure there are others, not many though. Politics does not work well with "most of these projects failed". It's too easy of a target. Anything that failed, is an easy target for a congressman to take aim at as be…

I’m not sure that’s really true.

The NIH budget is about $40B, 90 percent of which goes to research via about 50,000 extramural grants and the NIH’s own intramural research program. Very little of this is expected to “work” in the sense that it produces an immediate improvement in human health. Same goes for the NSF. The DoE, NASA, and DoD also fund plenty of research beyond the ARPAs (DARPA, IARPA, HSARPA, etc).

The ARPA programs are outliers in that they’re more likely to take larger risks in exchange for bigger rewards, while the NIH (etc) are more incremental. Nevertheless, it’s not true that these programs try to avoid ‘failures’ at all costs.

Post reply on HN