This will always be the shortcoming of how "knowledge work" sort of job interviews put so much focus on trying to get the exact unicorn candidate from the outset, instead of just seeing what happens and either keeping the worker or letting them go like how every other job works. Most shift work on the other hand doesn't have to worry about you lying on your resume or interview. If you lie, and you can't do the job the next day, you are fired. That's it. If you lie, and you can do the job the next day, all is well anyhow, right?
It is like an unhackable way to do it. Just hiring the person, seeing how it goes, and letting them go or not. It isn't rocket science. The guys digging a ditch outside the office know more about hiring than the career HR professionals inside I guess.
It isn't like he is the one who did the research to lead to that decision though. That just isn't his job. He isn't an analyst. He is CEO, who reads reports or sat in presentations made by his analysts, whatever their specific position titles, who are really the ones who did the work weighing and making this decision. His role was to say yes to these people whose job it is to actually figure this out.
So indeed, if he was not CEO, likely whoever would have been would be given the same information as Jensen that lead to him saying yes on tensor cores. And in all likelihood, the same exact decision would have been made, or at least a coinflips odds chance if it really were a matter of "yes" or "no" with no clear winner.
My experience in corporate America is that, no - it would NOT be likely that a different CEO makes the same call.
Many, especially the "non-founder" CEOs, are risk-averse.
There's likely many CUDA-like ideas from internal engineers shot-down or ignored by leadership at companies like Microsoft, Google, etc.
While Tim Cook certainly drove the valuation of Apple up during his tenure, one of the most common criticisms was that Apple stopped innovating and more recently their quality bar has been dropping.
The fact that it is always Big Even Number shows how stupid it is. Imagine if it was serious like "23.5176%" then you'd know they actually did their homework on leaning up operations instead of just cutting off limbs and seeing if it can still walk or not.
You and I and many of the commenters would be on there. People think these CEOs are wizards performing black secret magic we'd fail to understand.
No. They have the same 24 hour day as you and I. They think at the same speed and talk at the same speed. Their "workload" whatever it may be, is finite, evidently finite enough to juggle several romatic relationships at the same time, with much actual work and decisionmaking delegated and informed by domain experts who will be there no matter the CEO.
So yes, you or I could in fact be CEO of NVIDIA. It would be easy. Reports would be made for us by the worlds experts in the fields on what is going on. Departments would be ran by the best available talent on earth. My emails are being read and written by a dozen plus secretaries. Every presser I make I have an ivy league grad writing me the speech making me look like Hemingway out there. Every decision I'd make informed by a team of experts. Every plan thought of already by a hundred people smarter than me. All we'd have to do in the role as NVIDIA CEO is smile, waive, and read the lines handed to us on the sheet of paper by someone else. The ship runs itself even though we all pretend it's the CEO paddling 24/7 making it possible.
Really, that's all any of these guys are doing. Very similar to say the job of a pop star. Are you charismatic? Can you read lines? You'd make a great CEO then.
It would be a great thing for the economy. So many sectors are underinvested right now because capital has been stowed away into these companies valuations. Imagine 6T dumped in any other industry and it would be a generational step change event. For reference that's ballpark like 3x the valuation of the entire American biotechnology industry, all in one company. Talk about opportunity cost.
Suppose NVIDIA ceased to exist today. What would happen? AMD would step in. Huawei would step in. Maybe even Intel eventually. More fab capacity would go to these companies accelerators.
At worse, progress stalls for a year or two.
NVIDIA was at the lead and took most of the revenue and valuation because of it. But the same process would have happened with or without NVIDIA.
Now go back five years and substitute an NVIDIA collapse. Do you think that we'd be in a very different position for AI? I don't think it would be far off at all.
There were many game accelerator companies when NVIDIA started, they were just slightly better than the others, and the efficiency gains of consolidation resulted in those other companies being sold off.
ASML is definitely a one off though. And perhaps TSMC is a unique one off, but they got there by taking the most customers. If it wasn't them, then somebody else would have done it.
The key selection function was "this is the best option at the time" and if the best option hasn't existed, somebody else would have been chosen path that time, giving that other company the capital to build somethjng.
In addition to all that, other investments would be made in other sectors. I'm sure a huge bulk of investment into NVDA in particular is strictly because of the five figure percentage growth it has enjoyed over the last 6 years after all, with the whole chip argument being the fig leaf over the rampant greedy pumping. It has exploded like some lucky shitcoin and that's enough to drive investor action on massive scale after all. We are in the world where shitcoins are listed along with the blue chip tickers on bloomberg news. Everyone is greedy right now. Gambling on every scale from every direction. Most corrupt president in history to boot openly insider trading; water is full of chum and blood right now.
More people, more awareness about how money works, and more dependence on global investors.
Everyone invests so they can have more in the future, so yes, it's not crazy to imagine people will invest in a company that's growing without knowing any other details.
Why did it grow so large is the question. Why do some of these companies attract all the talent and grow so large and dominate the future is the question.
It's because we're all limited on time and greedy.
Valuation has no connection to “talent”. Do you really think NVIDIA has just ultra amazing employees that no one else does? No. It’s all smoke, mirrors and financial shenanigans. That’s not “talent”, it’s lack of scruples. For an even more obvious example look at Musk’s companies where the valuations have absolutely no connection whatsoever to the business fundamentals.
In part, because money is going to these other ventures instead. Where would money go if there weren't an NVDA is the 6T question. What would this world look like with more balanced investment across the SP500 sectors instead of money dumped into this one industry in particular? I don't think it would be more chip companies on a 1:1 basis of NVDA investment. The fact that it is going to a single ticker in this case makes it more likely to suck the other options out of existence. If it were say 100 chip companies, more likely on an individual level than an investor might instead say, maybe this time I would invest in some other industry. Mutual funds that investors buy into to hedge into many sectors are also heavier than ever in tech companies like NVDA sort of perverting the whole concept of these funds in the first place. Sort of unprecedented times between how overlevereged things are, and the constant uninformed pumping of air into these stocks thanks to these packages such as these mutual funds that people automatically contribute their paycheck into, again believing they are probably getting much more broad exposure vs how leveraged these packages are currently.
He gave his employees some of the wealth. And his employees seem to like him, or at least they did a few days ago. Maybe he's not selfless but he certainly wasn't completely selfish.
There must be something redeemable about him that makes people want to work for him. That's all I meant by selfless.
Company valuations aren't a sum of the cash flows invested into a company or the available cash flows out of liquidating it either. These aren't uncorrelated numbers but you can technically get a six trillion dollar "valuation" from one person paying $1 for a sixth of a trillionth of a company.
No they aren't, but it is an example of relative scale of investment. Especially when much of that market cap growth has come just within the last 5-6 years or so. Money could have flowed to other ponds collecting off the river of investor cash flow instead of this one massive lake. This might have lead to a different growth trajectory in our economy and technological development than one so focused on the ROI of chip production in particular, to the exclusion of much else. The fact that much of that demand is brought up by circular investment is another aspect. Two lakes backfeeding into eachother eating more and more of the flow off cash flow river, drying up other ponds along the way. Much is lost with such unitary focus of investment.
I would go a little bit further and just say that American workers haven’t been the main beneficiaries of investment for at least a decade now. It “used to be” (some rose tinted glasses but still big kernel of truth) that if a company was making a lot of money or sitting on a big cash war chest, eventually it would make its way back to the employees for expansion, R&D, etc. It would be used for something! But ever since companies started to divert those flows to stock buybacks (and now, the equivalent which seems to be mutual stock purchases) less and less of those profits have been functionally reinvested - the reinvestment mostly happens on paper, and the already-wealthy owners of the stocks get wealthier without expanding the stockholder base as much.
It would be really interesting to get a snapshot of the actual dollars spent on ownership at any one point in time. Stock market volume can be a lot of froth and slosh of a small-ish subset of all shares.
Every share purchased on the secondary market is sold by someone else on the secondary market, you can measure the net cash flow between companies and equity investors by tracking issuances on the primary market, dividends, and stock buybacks.
Does it work like that though? I agree that AI may be crowding out other industries, but I'm not sure you can just take apart a company and redistribute that money elsewhere.
No, it isn't measuring the velocity of incoming investment. But it goes to show just how outstandingly large the pile of money on one ticker in particular is, relative to other high tech industries that are even favored by investors for their 10x potential (maybe moreso historically than today albeit).
Older than the internet. On the crumbling ruin covered in goat shit reads "My name is Ozymandias, king of kings: Look on my works, ye Mighty, and despair!"
It was pretty much useless anyhow. Just some "power users" asking if you've done the 3 steps that aren't going to fix the issue from some apple support article telling you to turn it off and on.
If the fix was anything under the hood beyond a dedicated system settings toggle in the gui, it was not going to be mentioned or discussed in this forum. You weren't going to get your launchd questions solved here.
Only for the most boilerplate issues I would find. Actual technical discussion best I've seen is github issues. Finally here are the users who know how to peak behind the curtain of the gui and we can go right to exchanging patched forks ahead of even the maintainer pushing a change.
A lot of lottery players will say use a birthday or something. Kind of interesting to think about this as deciding on pure random search or setting a seed and sampling that seed over the space. The behavior becomes much less about juju and maybe more an actual functional one with some biological utility, this ability to sample a defined input across a function set. The lottery aspect is almost blinding the point of why it might be useful to think like that.
About (dear lord) 30 years ago when the national lottery came to the shores of Britain, my family had the neighbours over and there was some faux hand holding "seance/trance" make-pretend, with us spitting out numbers for the very first lottery ticket. We won a tenner!
Silliness of the highest order. I still have those numbers written down. They've got the same chance as any other set - but this set are _mine_. I might just be the only person alive who was there, now I reflect upon it. Huh
> A lot of lottery players will say use a birthday or something.
Ironically, this makes the ticket less valuable.
You can't do anything to change the probability of winning, but you can change the probability of winning at the same time as other players (which affects the size of the winnings). By picking birthdays (i.e. forcing the numbers to be in the first 30 or so), your prize, if you do win, is likely to be smaller than a really random selection.
Guitar music in particular has very little actual pressure on learning any music notation or theory. Some of the greatest guitar soloists of all time could not read sheet music and knew zero music theory. This is common enough where there's good odds if you just name a great guitarist, they are probably in this camp.
This is debatable because I dont know what you mean when you say they didnt know theory. Hendrix is a great example, supposedly he didnt know, but it is very clear he would play the pentatonic of the quality of chord he was playing over. Is that knowing theory? Or playing is playing the BB King box theory? Or is knowing the major scale theory? It's almost certain that the best guitarists that you're saying knew some of these things. So do you mean the nomenclature?
Like, I know what a secondary dominant is because I've had to use it in blues turn arounds- I didn't know that's what some folks would call it, but I knew that in the key of G, I'd often play an A instead of the Am.
And I think that the vocabulary should describe what folks do rather than the doings being driven by the vocabulary.
Most guitar players (and improv folks) that I know have much more theory that the folks who are reading but who generally don't think about what the function of their line is in a larger song.
If you're playing by memory and ear, you have to have some concepts because you can't just read a part.
I've seen it and I kind of get it. I have worked with folks who do things for theoretical reasons instead of describing what they do with these ideas.
At the same time, I wish more folks would learn to read music, or even be literate enough to read lead sheets- my life would be so much easier if folks could just show up and play the thing without spending hours learning from recordings and making whatever kind of esoteric notations they've made up.
In Miles Davis' autobiography, Davis talked about how they jammed a few times. Davis would tell him to play a particular chord or scale, and Hendrix had no idea what it was at first, e.g. "try an E minor diminished yada yada there... a mixolydian scale would work on top of these changes..."
Once Davis taught him the concepts he got them immediately, or it might be a chord he knew and played, but didn't know the formal name of. So, he was a brilliant intuitive musician, but didn't have the theory chops of someone who attended Juliard like Davis.
That is what we mean by Hendrix didn't know the theory.
Yeah he didn't know the names of the concepts that he already knew. My intuition says music theory is the understanding of the concepts, not the formal study of the names. Like a person that can only play the major scale because they studied the theory, in my mind they have less mastery of theory than a person that can compose "Little Wing" or "Lenny" without ever studying it.
To me, the theory is as much a communication tool as a knowledge. It is a common language for musicians to communicate musical ideas, at a higher level than musical notation. E.g. I can go to a jazz open mic, someone says the progression is "ii V I in Dminor", or a blues jam and its "I IV V, A major", or a Grateful Dead-type jam and its all mixolydian in B, and everyone is on the same page out of the gate. The theory is you know all of the scale modes, AND what they are called, AND can tell another musician what you're doing and why. Tons of people can play interesting things outside of the box, without know why or how it's defined.
I can get behind this definition. But even following that definition, people without theory knowledge who know how to play would know a third of theory if you split the three categories that you said equally. I venture to say that the concepts themselves weigh more than an even third though.
Right, you can be an incredibly articulate speaker without necessarily knowing the grammatical terms for the different parts of speech in each sentence.
I think there's a number of people who learned to play by ear who just sort of "internalized" theory without needing a formal curriculum.
They have taste for sure and an ear for a blues lick, but they don't know formal theory. It's like how my cat understands gravity but has no awareness of formal gravitational theory or the maths involved.
The cat is a bad analogy. A better one for me would be like a mathematician that does a contribution to the field without formal education. Would you say they don't know any mathematics if they invented their own nomenclature?
Well it isn't like they are inventing their own nomenclature either. Where are Jimi's scale books he wrote in his own modes? He didn't operate in those terms. He was very much like my cat, having an insinct for a formalized theory he does not know the formalized language for.
Is it understanding formal music theory? No, only if you stretch the definition of formal music theory to include what is definitely not formal music theory by any one's definition.
Look at the tablature for little wing and you will see that he most definitely did some sort of theorizing, as he was very structured and systematic.
My point is that what he did is what music theory describes, and he definitely arrived there. He most definitely could play the scales and the modes, even if he never used the word dorian.
There is just some allergic reaction to the word that people take to mean something along the lines of demeriting the artistry. Had he learned the diatonic chords would that mean he is no longer a no theory guy, and would that make him worse?
I know thats not what you are saying, but i do see that from the hardcore 'theory is bad' proponents. Is theory the names of the relationship between tones, or the relationships themselves?
He had a music theory I'm sure, but the overlap of that with what's currently being taught in conservatories (let alone what was taught in his time), I'm less sure about.
I'm not really tracking where your confusion is. You can very much so do or "discover" something in practice that a theory describes without knowing that theory. I'd say it's even very common to "know" certain notes/chords/whatnot sound good with others and not learn "why" until later on, if ever.
My confusion is: why would you say that they don't know theory if they know the thing that the theory describes? If a person builds a radio by themselves, would you say that they don't know engineering because they didn't get a formal education on it? If a person figures out how to make a derivative, would you say they don't know mathematics because they didn't study it formally?
Anyone? I have a hard time finding the beat. If you ask me to count some 4/4 song, there's only about a 25% chance I get the one correct...
I'm a very beginner guitar player (and have been for about 30 years now) and I know a couple of simple songs. If I try to play along with a record, I'm fine until vocals kick in and then I instantly lose track of the beat and find myself strumming to the syllables of whatever the vocalist is singing.
I agree. But the theory is a way of structuring a system of names and shorthands to help codify what you know about producing music. At what point does your system for structuring the understanding you have about how to organize pitches in a way that you like become theory? Specially if it matches one for one to established concepts in academic music theory?
On some level, yes, knowing what shape an G chord is on a guitar is knowing some theory in the exact same way that being able to get hello world functioning in a language is necessary to starting to understand the broader elements of computer science.
Just because something is the Nth lowest first step does not mean it's not a step.
The argument was that the greatest guitarists didn't know theory so it doesn't equate to that. Hello world equivalent. It's more like building TempleOS or Rollercoaster Tycoon or building SQLite without getting a formal education somehow means you don't understand computer science
It's just really all about the ear and there's plenty of fabulous guitar players who will tell you to transcribe Charlie Parker but tell you don't bother playing scales for hours a day
the only thing id clarify is that a lot of them won’t _formally_ know music theory, but their playing demonstrates can demonstrate a deep understanding of it.
Music notation on a 5-line staff is just another in the category of "shapes," along with pentatonic scales on a fretboard. Theory is another matter though.
Honestly, I know a ton of folks sitting 5th chair violin in a community orchestry or playing 3rd trumpet in a big band who have no idea what note in the chord they are playing and just reading and repeating the line.
Most of the musicians I know have very minimal theory, ironically even less than folks who can't read.
It is like an unhackable way to do it. Just hiring the person, seeing how it goes, and letting them go or not. It isn't rocket science. The guys digging a ditch outside the office know more about hiring than the career HR professionals inside I guess.
reply