741 karma · joined October 6, 2021
I don’t think most people looking to build an AI company want to build an LLM and call it a company.
I don’t know if this phenomenon is aptly characterized as “too much efficiency”.
This point is at least contentious. I am not going to go into weeds, but I fail to see how you can make a product that literally popularizes the term “network effects” without vision. You can say he “stole” the idea for facebook, but what about everything after? Ideas are cheap.
> no ability to execute
I see no evidence of this. How does someone scale from $0 to $1T without the ability to execute?
> Spending $50B in a few years with almost no revenue to show for it has to be the biggest business failure of all time.
I also think the metaverse was a failure. But I think they failed fast and have more experience in that area than any other company. They basically spent $50B to get data on the intial vision for the metaverse. Their stock and business has fully recovered.
It was a hedge against being “too late” and they got the timing wrong.
It’s one thing if you fail and the business goes bankrupt. It’s a completely different scenario when they recover completely. How does “the biggest business failure of all time” not end in that business going bankrupt?
Off the top of my head I am pretty sure hundreds of billions of dollars of value vanished after the Lehman Brothers collapse.
In tech (adjacent) you got worldcom, which was of a similar magnitude but in y2k dollars.
Generally, a large magnitude of value is created during "platform shifts" and they have now placed (really good) bets on VR/AR and AI (LLMs).
If you get a demo of someone editing stuff super quickly using only keybinds I think it would look pretty cool.
Right now I don’t know if the value prop is strong enough for me to switch editors and remodel/relearn all my keybinds. This is like a 10-15% improvement not a 1.5x or 2x improvement.
An extension would probably help onboard users.
There is way too much stuff about all the meta around making projects and just plain clout chasing rather than sharing intellectually interesting projects. I had twitter for an hour before I deleted it because I realized it was just a big popularity contest. The SNR was just too low.
I still think "building in public" is a good thing apart from the buzzword-y semantics it has taken on. The best way to do this is to talk only about the project and the technical challenges it has, and view "building in public" as a moral commitment rather than a marketing one. Perhaps "moral" is too strong a word. I really mean sharing things, not to boost your ego or flex status, but because you think it's actually cool/useful.
It's not really a job people should be hiring/applying for.
Why? I really couldn't say. I think we just like the feel of it. The only reason I type with proper capitalization on HN and my blog is because I know older people read it.
As someone who just played the college admissions game last year, these (US News) rankings pretty accurately reflect the *perceptions* of the general, college applying, public (or maybe they are the source of those perceptions). They aren't really even good at that outside the T10.
In bigger companies, it's a zero sum game. They don't really care about you because their scale makes it hard to identify who cares for them, so everything is just a business transaction.
The let "let boys be childish" part and the overall psuedo-human tone kind of alarmed me. The random "hahas" littered around, seemed like a robot trying to be a human.
> micromanaging
He has a playbook/formula that works and all employees are solely focused on executing that vision. People have little operational ownership. In other words, employees don't have freedom in vision.
I even said it probably is necessary for the success of his business that employees don't have that freedom. I just would not enjoy working in a environment like that and I think employees (especially early ones) need to have that kind of operational freedom in startups (which is the context of my comment).
I would say there is a mistreatment and reductionistic mischaracterization of the current age as a “void”. Further, the choice of paganism (belief) and the “void” is a false dichotomy. Post-structuralism partially developed as a response to the “void”.
I don’t think you can faithfully engage with post-structural ideas and simply write it off as nonsense—they seem to answer a lot of the questions in the article.
Particularly the part where the author suggests atheists form a sort of pseudo-religion. I think post-structuralism deals with how to operate when left only with belief and narrative (i.e. only the pseudo-pagan performance). The conclusions in the article seem very similar to post-structural first principles/common strands of thinking.
Of course, you could write off those post-structural ideas as non-sense because it attempts to make sense of nothing (literally, the “void”). Sort of a garbage in garbage out scenario.
I could probably write a whole essay in response, so to cut it short I just said “post-structural ideas”, which, in fairness, is broad, generally a very loaded term, and in need of some more elaboration.
The difference is the game he's playing (youtube) is similar to the game we're playing (startups) so the success is tantamount.
The game tobacco companies play is also very different, so the tobacco companies success will teach you very little about being successful in startups.
If a startup can't attract talent (a sign of bad traction), that startup probably is not that good and more people won't solve the underlying problem. You would also be surprised how many startups outsource dev/marketing/etc. in their initial stages.
If you can't convince smart people to work for you and that your idea is good, good luck trying to convince customers of the same.
>And that 95% of startups don't know their metrics. Pretty sure almost all do but again don't have the skills or resources to meaningfully move them.
I said most don't know them as well as Mr. Beast. Read "Chapter 1: What makes a Youtube video viral?". Most founders have not put the same amount of time into seeing how to track, measure, and impact metrics. He identified key KPIs and then experimented with changes until he found what worked. His whole north star to, minute by minute, structure each video, is informed by the KPIs. His whole strategy is built upon metrics by metrics.
He clearly is obsessed with them to a degree few are. Some startups don't even know how much money they make, how much money they lose, etc.
1. KPIs, for Beast they are CTR, AVD, AVP, will look different if you are a startup. I am willing to bet he knows his metrics better than >95% of startup founders. Because he is literally hacking/being judged by an algorithm, his KPIs will matter more and can be closely dissected. Startups aren't that easy in that sense, but KPIs still matter.
2. Hiring only A-players. Bloated teams kill startups.
3. Building value > making money
4. Rewarding employees who make value for the business and think like founders/equity owners, not employees.
5. Understanding that some videos only his team can do, and actively exploiting and widening that gap.
The management/communication stuff is mostly about working on set/dealing with physical scale. You need a lot more hands dealing with logistics, which requires hardline communication and management. In startups, the team is usually really lean and technical, so management becomes more straightforward.
I am also getting some bad culture vibes from the PDF and really dislike the writing style. I think it's important not to micromanage to the extent he is--it's necessary, maybe, for his business. Not for startups. Interesting perspective, reminds me of a chef de cuisine in a cutthroat 90s kitchen. The dishes (videos) have to be perfect, they require a lot of prep and a lot of hands, and you have to consistently pump them out.
The rest I chose are examples from YC as they are the ones with phrase “Make something people want”.
I think if I spent some more time I could have came up with better ones.
EDIT: On the other hand, all external stimuli changes the way we think, we are meaning making machines. Where there is something we try and make meaning of it.
I think scale matters in terms of things impact on our thinking, but fundamentally the argument could be made that for all things people want there are changes in the way they think.
Maybe clarifying “novel things” people want would be clearer.
In my mind the mapping and inverse mapping were defined in these two paragraphs:
“I think altering thinking at scale requires people wanting that thing, otherwise you wouldn’t get network effects and scaling would be impossible.
Also, I think that making something people wants requires altering the way they think. When you make something new you ask people to reconsider their existing habits and first principles. This mental shift, however small, is what leads to mass adoption and users falling in love with your product.”
I tend to disagree. I have a few hundred friends/people I follow on Instagram and I see people post stories and posts about their life all the time. People are generally trying to portray "Look at me! I have such a cool life!". This is how it is for young people.
As for TikTok, I find myself scrolling with no end in sight. I think the studies on the addiction level of shortform content and its long term effects is going to be extremely shocking.
I can watch an episode or two of a show, maybe 30m-1hr, and turn it off. But TikTok keeps me constantly stimulated and I can easily forget what I am watching and "doomscroll" for hours. With TV I make the conscious decision to watch it. With TikTok I have to make the conscious decision to not watch it or open the app. This may be a side effect of how my brain works, but a lot of young people have similar experiences.
I think the main difference in our experiences is a generational gap in how social media is used.
In 2017 I thought I missed bitcoin, still managed to mine a meager amount on my parents computers. In the modern day I was proven wrong.
There is something to be said about doing things "the hard way" and I think AI is, in the short term, going to decrease the amount of medium-skilled software developers out there.
AI (LLMs) represents a platform shift on a large scale and I feel the environment is similar to that which birthed Zuckerberg type entrepreneurs. I think in the next 5 years Gen Z will have its heyday in terms of tech startups... At least, that's what I tell myself.
Yeah, I agree the Web3.0 stuff was overhyped--not really a platform or thinking shift. There are a lot of black-hat hackers who are Gen Z, because generally only young people take risks like that while simultaneously having the skills to get paid a tech salary.