Someone Has to Run the Fabs
noahpinion.blog
noahpinion.blog
Perhaps the best way to train the next generation is to create actual bootcamps on-site where young professionals get a chance to try and answer questions like "How did lot ABC123 end up with unexpected CU exposure while it was sitting inside a N2-purged storage area?" The only tools they get to use: historical logs of every event that occurred and a copy of grep.
In my experience, the deep PhD-style education was absolutely not essential for the day-to-day. Only a few people need to master the idea. Once the complex academia is converted into a physical reticle or CMP process recipe, you then need an army of people to reliably execute on that standardized vision.
I think someone coming into semiconductor manufacturing would have an easier time if they were dropped right into it on day one. The theory is nice but we found it to be mostly a waste of time when someone from manufacturing was calling us on the phone and complaining about a stuck lot.
In other words, apprenticeship. I strongly agree, there should be much more of it in our industries.
This is the case: https://en.wikipedia.org/wiki/Griggs_v._Duke_Power_Co.
Since then (whether you agree with the ruling or not), companies have been reticent to use aptitude tests. Instead they outsource selection to colleges, which conveniently do the exact same thing as the SC was trying to end. Except now, the companies can foist the cost of training onto the employee, since the expectation was no longer that you could get a job out of high school and learn on the job.
Unfortunately, the case IMO has had deleterious effects. It should have been expected, given the racial dynamics of the time, that any aptitude test may have produced unequal results. However, the proper response is the government improving the education system so that these disparities disappeared, rather than making them illegal.
As this article itself states, if an equal mathematics aptitude test were run today by a company seeking to train up engineers and other STEM-inclined workers, then we would expect the test to produce racially disparate results (because according to the article Asian and white parents, who are better off, invested in private tutors for their children). No company will do this test then, because it will certainly be found to be discriminatory once they get the results back, and suddenly they're potentially on the hook for millions. Thus, companies instead outsource the selection to colleges who suddenly manage to whitewash the fact that the graduates of their STEM programs are disproportionately asian (and to a lesser extent, white).
But this is why apprenticeships are all but impossible in the US. Without the college degree to exculpate you, you will be certainly be found to have discriminatory hiring practices if you are picky with your entrants.
There's a book I like called "Stop Guessing" (Nat Greene) about that.
Here's the US Amazon link: https://www.amazon.com/Stop-Guessing-Behaviors-Problem-Solve...
Not affiliated.
Unfortunately the US approach has been to throw out every research study suggesting it might work, while failing to fund better studies. See https://ies.ed.gov/ncee/wwc/Docs/InterventionReports/wwc_sin.... Behind this is that fact that the educational establishment makes a ton of money off of rewriting textbooks and retraining teachers for each reform wave. The hasty changes are guaranteed to lead to disaster, which motivates the NEXT reform wave, which they then also profit on. My mother became a school teacher in the 1950s, and this cycle was ALREADY in progress back then. (Think Sputnik leads to New Math leads to Why Johnny Can't Add, etc.)
Switching to already existing books for an approach that actually will work would be a disaster for the education establishment. And so we'll never do it. No matter how embarrassing our regular (but profitable for the education establishment) disasters are.
I complain about this occasionally. See https://news.ycombinator.com/item?id=13822577 for a more general complaint, and https://news.ycombinator.com/item?id=34634210 for a rant about the Common Core disaster specificially.
Housing shortages are profitable to property owners, bad rail infrastructure is profitable to airlines, complicated tax codes are profitable for accountants, bad foreign policy is profitable for defense contractors, etc.
All complex societies suffer from this but it seems particularly bad in the USA for some reason.
So, most of the positive changes we make come from either random whims of the rich: "we can't have that here, my sinuses are too delicate", or moments of revolt that scare the rich into action. But the actions also tend to be random and often go down the authoritarian maintain-status-quo route.
The combination of "this problem has clear winners" and "anyone can veto any change, but no one can veto the status quo" makes for entrenched, powerful interests.
I suspect that it is particularly bad in the US because our insane vetocracy.
I've seen the Russian School of Mathematics and I wonder if there are other "foreign" style of math education. If so, I would consider enrolling my future children.
When I grew up, I was enrolled in a bilingual school in middle-school. So I took math in Spanish (following the Mexican curriculum), and in English (following the US curriculum). I thought that was fantastic not because one or other was better (I can't recall) but because it gave different approaches to the same topics.
It also spreads each concept over time so that you keep revisiting the same ideas to ensure it gets fully integrated. And mixes them. So grade 1 kids may be doing "multiplication" by arranging blocks in a grid before they've fully mastered the + sign.
The USA gives a much more cursory pass through operational actions and goes straight to the abstract representation. We also do a lot more of one concept at a time, only to move on to the next while kids forget the first. The result is that kids learn to blindly follow poorly learned procedures, hoping to do it well enough to pass the test before forgetting the concept. And then the following year do it all over again, having forgot the material from last year.
Yeah, yeah, we have an official curriculum, sequencing, and there is lots of verbiage in official documents about how and when kids are supposed to get reminded, etc. But I described what ACTUALLY tends to happen. And you only have to glance at test scores to realize that the official narrative about what is learned is more myth than reality.
Every math concept had physical "materials" that we had to manipulate to get to the answer (e.g. for square root, you place N beads on a grid in a square form and count how many are on the side). Often, tricks to get quickly at the answer were discouraged early on to promote a proper "grasp" on what the operations entail.
His thinking at least parallels Marie Montessori's. There probably is some direct influence, but I couldn't find evidence of it in a quick search.
However it does not surprise me that the approaches would work out to be similar.
Now I live in a family of nerds amidst a sea of extended family who are all rather poorly educated. I watched as my cousins cared more about ski-doos and video games than they did about their math homework. Unsurprisingly, they didn't get far academically and now as they are approaching their adult life, are struggling to be employed and do not like their job prospects, the same as their parents did before them.
Meanwhile, me and my siblings were eagerly pushed to pursue education. Curiosity was ignited at every opportunity to encourage us to think hard and explore big ideas. This has paid off in spades for our family. Innovation, thinking, and exploration are often pre-cursors to success.
Seeing my extended family (and their friends) justify their uneducated views of the world, hearing their bigotry, listening to their hatred against "smart people". It hurts my soul. Knowing that my extended family out numbers my close family 5 to 1 by family size makes me worry about the future of our society.
These ignorant folks are voting with their wallets, ballots, and social media voices. What does that say about the future of North America? What does it say about the changes we need to make to enlighten our people for the future?
i honestly thought this had to be satire - it seems it wasn't.
Kids are failing? Lower the bar! Suddenly everyone has "improved". The bureaucrats are happy, the unions are happy, even the parents who don't care about education are happy!
Makes it easier to push for the next part of the bureaucrat's agenda [0] [1] : CRT laced math and computer science education.
[0] https://equitablemath.org/wp-content/uploads/sites/2/2020/11...
but people should have the chance of being taught these things, or indeed anything. to deny them of this is flatly wrong.
We built an actively knowledge-hostile culture that glorifies and rewards managers, hustlers, grifters, and influencers, and we're then surprised that nobody with ambition wants to... Actually build things.
Also, I mentally separate long working hours from other aspects of “worker conditions.” You can have low working hours but a toxic and unsafe work environment or a physically safe and supportive work environment with long hours. You can also have high and low pay with various working hours (although typically high pay isn’t supported by low working hours unless the business has some sort of massive moat).
Of course it's possible that insane working hours were why TSMC employees got EUV photolithography working at scale before any American manufacturer did. Was that the key difference between Intel and TSMC? It's a question that I personally haven't seen an answer to.
This wouldn’t be a problem, we just let Taiwan work themselves silly why we share in the benefit by getting better and better computer chips, but that doesn’t work if we want to COMPETE with Taiwan in this area.
Educated people are less likely to swallow corporate and Government bullshit and more difficult to control. Companies call them "overqualified".
As in, you might be willing to build, but soon you might be frustrated that you don't see much of the fruits of your labor in terms of mobility or rewards.
The pushback against the capitalist class is real and growing amongst the younger generations.
It helps when your society is covered in frescoes and statues that glorify the engineer and the bricklayer...
But when the wastrel of a director is driving a BMW, and you have a ten-year waiting list for a broken-down Lada, you may still feel some resentment against said director.
The 'Rewards' side of things is pretty decent in some parts of STEM, but are rather lousy in others. Hardware work does poorly on this scale.
Americans need to readjust their attitude toward math. People just say “oh I’m not good at math” because nobody will shame you for not being good at math. It’s not considered a personal flaw. There’s no social pressure for the middle/upper middle part of the workforce to bang their heads against it and get good as they can at math.
My undergrad degree was in math, but neither my wife nor either of our (now-adult) kids think they're any good at it. They smile tolerantly when I sometimes shake my head over their seeming inability to grasp — and their evident lack of interest in — what, to me, are simple mathematical concepts.
(Of course, my own higher-math skills have atrophied to nearly nothing after decades of not using them ....)
Can you explain the relationship between anti-immigration and manufacturing success? As the sibling comments says, the USA has far more immigration than, well, anywhere in the world.
By total numbers, probably. As a % of population, I don't think the US comes even close to the top. Probably even Canada and other North American countries have a even higher percentage, so it doesn't even have the most in North America.
And just eyeballing it, you'll notice most countries demographics don't radically change in the span of a generation.
https://en.wikipedia.org/wiki/List_of_sovereign_states_and_d...
I’m not sure about that compared to other countries, but I’m also not sure why it’d be desirable to have a higher percentage. Seems like it’d be extremely risky to an existing society to even hit something like 33%
It has 10x the US per-capita immigration quotas. And it takes less than a single electoral cycle to go from permanent resident to citizen!
Unsurprisingly, the current liberal party in power is very popular among immigrant born population...
Immigration is not just about importing people. It's about importing a culture that is better suited for the demands of your society's ambitions. On the surface it's about bringing in people but really it's about bringing in a people.
Nope, America largely exterminated its native cultures hundreds of years ago. The culture that exists today is the immigrant culture of nation founded on ideas (freedom, slavery, capitalism, democracy) not bloodlines. America can be for anyone who wants to be part of that story.
I'm all for teaching arithmetic, but emphasizing the algorithms and how to prove them correct should be paramount. Teaching arithmetic in various bases is important, as should be the basics of numeric analysis. Formal logic should also be a core math subject.
But no, the USA has to teach subjects as were common in the 1920-1950 time frame, in the same manner as back then. This is a consequence of "home rule".
Memorizing facts is the foundation of all learning. When you add 7+7 in your head do you instantly know the answer is 14 because of "understanding" or because you memorized over all the years.
Also the current US education system strictly deemphasizes memorization. Instead it pushes for "understanding" (conveniently a concept that is not measurable).
> Like anybody doesn't have multiple calculators in their pockets.
Ok and?
China and India are producing the smartest people in the world and their education systems strongly emphasize memorization. Understanding comes when you've memorized enough information that you can connect the dots.
Once you get to advanced math it's basically impossible to understand from the start and you can only rely on memorization until you've memorized enough and experienced enough that it finally clicks.
The best mathematicians aren't the ones just proving theorems anymore, they're discovering interesting conjectures through exploration and then handing the grunt work off to a computer. Engineers aren't calculating anything remotely complex by hand anymore, nor should they. The best scientists also tend not to be people who were problem set whizzes in school.
I'm sorry no. This is simply wrong. Computer proofs are not there yet, and plus, a lot of the interesting theorems (including many of the well-known ones) are non-constructive which are difficult to prove on a computer. In fact, many theorem provers cannot prove 'basic' facts about math due to how constructive proof theory works. Indeed it is not clear what truth means at that level, but considering we regularly use non-constructive proofs to build things, I'm not sure that matters.
I think what is most undervalued is the utility of sheer competence and technical skill to creativity. Without the former, there is no latter. If someone who never studied piano told you they had a beautiful piano solo in their head, would you believe them? More importantly, would you trust them to execute on the creative impulse?
If a person who doesn't play piano told me they have a piano solo in their head, i would set them up with a midi sequencer so they can bring it to life. My preconceptions about their capabilities are irrelevant.
A MIDI sequencer is not a piano, and mathematicians typically don't use maple/mathematica/octave/python (unless they're in applied maths, which is a subset of mathematics as a whole).
The most important breakthroughs in mathematics, the kinds of things that push you over the edge as a society and that expand the scope of humanity's knowledge, are at the proof level, which those programs cannot achieve.
Your MIDI sequencer analogy is very apt, because while someone may be able to bang something out on the MIDI sequencer and have the computer play it, a MIDI sequencing is not a human piano performance.
But you're right, if you want a hollow fascimile of the real thing, then yes, those tools could work to impress someone.
I’m no mathematician, but I’d be very surprised if the overlap these tools have with mathematical research is very large.
Yeah, no disagreement there.
> The last attempt to fix it (common core) mostly emphasized memorizing arithmetic facts.
In fact, just the opposite. The old math system was rote memorization. Then common core came along and made all those parents and their limited math skills obsolete. It didn't help that teachers who also learned the old math couldn't or didn't understand common core methodology.
BTW, I'm a parent of one those school kids who was trying to learn common core. I got it right away and don't consider myself particularly brilliant. But the grumbling from so many parents who just thought they knew more than the educators was deafening. In the end, the state dropped common core and they are still complaining about it years later. The right made it into a boogeyman and now that and trans kids are all they can fixate on.
No arguments there.
> The last attempt to fix it (common core) mostly emphasized memorizing arithmetic facts
Did it? My understanding is that it uses a few different approaches to build basic intuition about numbers, and that parents hate it because many of those approaches don't make sense to them, because that's not how they learnt math.
Which makes it an incredibly convenient punching bag.
In PRACTICE it was kids being present with a confusing array of things all at once when they hadn't sorted out the basics, and as a result learning none of them. See https://news.ycombinator.com/item?id=34634210 for my rant about what my son actually went through.
There is an old saying about writing. "Writing is easy, just open a vein and bleed." That particular rant I'm pretty sure that I hit an artery.
Unfortunately, it's a valid criticism.
More unfortunately, switching to another standard for teaching math isn't going to solve this.
[1] Why be society's punching bag, on the front-lines of an insane litigation and culture war, have your heavily-backloaded pay be contingent on putting up with decades of this kind of bullshit, all while trying to do a mixture of teaching and social work?
However I stand by my point that Common Core should be described in terms of what it did do, and not what its supporters hoped it would do.
Teaching requirements are NOT standardized at the federal level, and lots of states have decided to just lower standards instead of paying professionals more. If you only hire baby-sitters, don't be surprised when they can't teach simple stuff. In my state, we have high standards for teachers, and every single one, from kindergarten right through high school, from french teacher to cooking to history, could teach a middle school math course.
Your state or municipality cheaping out on your child's education is not the fault of federal regulation, or of a federal curriculum. You'd be whining about the same thing no matter what the national curriculum was, because the problem is that your local governments don't care about education.
https://www.nbcnews.com/news/world/rnc-embraces-common-core-...
> If you're familiar with Glenn Beck's broadcasts, you've no doubt heard about his unhinged crusade against Common Core, including his declaration last week that he will no longer send his kids to college -- Common Core will only indoctrinate them and make them part of "the system that is coming."
> Republicans are taking this all very seriously, with lawmakers in 18 states considering legislation to block Common Core, and Texas Gov. Rick Perry (R) denouncing it on Beck's radio show.
As mentioned in the article there are some actual experts with criticisms to make but many of the people getting unusually upset about it had got sucked into a conspiracy theory.
It's failing stem entirely from implementation and execution (on every level), leaving a lot of people wondering if general standards top-down are even the solution here, seeing how No Child Left Behind was handled.
We often mistake parental investment for 'in-born talent', while both are due to the lottery of birth, calling it 'in-born talent' makes it seem unachievable for others. It's a harder hill for sure, but not insurmountable.
I've taught many people to juggle from literally zero juggling experience. Hand eye coordination is not equally distributed among the population, and some people are naturally significantly more talented (and make more progress much more quickly than others. 10X differences are not a joke). I'm not sure why we keep believing intellectual things are any different.
Having gone to school with some exceptionally smart/not-smart people at both questionable public schools and then one of the best universities, the discrepancies there are definitely not just due to parental investment (though I'm not denying that it can play a role)
I think you'll find enough nerds to fill the engineering/science pipeline. The real problem is with the broad technical workforce, that needs to be bolstered.
So you don't think STEM education is important? It seems like you left that out of the list intentionally, but not sure.
If so, also consider not whether it is important to the fabs, but whether it is important to kids who eventually want and need jobs. You probably will be able to kill fabs with the nerds in the engineering/science pipeline, but you are then excluding 95% of the current student base.
100% agree! We need myriad education options for myriad types of students. But the legit critique of the "equity approach", as is happening in the Bay Area, is that it is you they aren't allowing students to advance in STEM (especially M), even students want to. It is not forcing a minimum level, it is capping max level at a relatively low/mid spot.
> they are doing even more than 12-hour days often
Running the fabs sounds like a shitty job for chumps. Stick to node.js.