Ludlum is rare among equipment manufacturers in that they design with user-servicability in mind. They happily took phone calls and emails from my 16-year-old self about how to diagnose and repair various instruments of theirs I had picked up on eBay, many of which were long obsolete. Their schematics are still open, to my knowledge.
I have no relation to the company, just a very satisfied customer.
This sounds more like proactive outreach to PIs based on AI-automated market research rather than direct AI-based gatekeeping by the granting agencies.
There’s certainly a case to be made about using LLMs to find needles in haystacks, since most grants tend to be awarded to “repeat offenders” rather than newcomers and outsiders* with different methodologies.
The irony here is that if you look at the job postings of quantum hardware vendors, they ask for a laundry list of skills that only a small handful of people on Earth realistically possess (you included).
People are given the impression that there's this outsized demand for Qiskit jockeys, when in reality, what we're currently calling quantum computers are basically physics experiments with the cables cleaned up and hidden in a cabinet. The results you get from these things are tightly coupled to their hardware implementation, and you need people who can work, or at least think, up and down the full stack to get even scientifically useful results. Same goes for quantum sensors, networks, and other so-called adjacent technologies.
You could cut all of it and it wouldn’t solve the problems you listed. It would just create new ones by closing off a major valve to the country’s R&D pipeline.
Assuming you believe R&D is a worthwhile investment, what non-government alternatives would you propose for funding it?
Now do it for an applied physicist with a family and a mortgage!
In theory, the work I do sits in that valley of death between where the government funds uncertain things at a $1e5-$5e6 level and where private capital funds things with more certainty at the $1e7-$5e7 level. It’s easy to burn a lot of labor and equipment on dead ends before you know something will scale.
While I dream of unshackling my research programs from the bureaucracy of the institutions they’re tied to, I do very much enjoy having an actual budget, paycheck, and health insurance.
With all the chaos currently befalling the academic research community, I’m curious what alternative models exist, or could potentially emerge, to support independent research at the $500k - $5m level.
I’m no expert in formal methods, but of all the tools I’ve used, Athena [0] has been one of my favorites to work with. The proofs read much more like what you’re used to seeing in math literature.
No, the non-exception is not absolutely necessary, and there are plenty of people on my staff who fit the description. There are also plenty more who Dunning-Krueger their way into thinking they do, but break down when challenged to do anything novel. Understand your options and choose your program carefully.
> Musk has shown an ability to make an impact in multiple fields for which he seems quite under qualified for, for which he did not have "significant intellectual preparation and tacit knowledge". He read alot.
He also had a giant pile of money from his PayPal windfall to hire the right people with the tacit knowledge to act on his ideas. The difference between a crank and eccentric businessman is the size of the budget they can wield when nobody else will.
> Academics are rarely comfortable discussing the shortcomings of academia.
Correct, which is why I’m not in academia.
Celebrity exceptions are exactly that; exceptions. Those people knew an opportunity when they had one, and were able to generalize their early successes into other domains by leveraging the financial, social, and intellectual capital they accumulated. People who fit this description aren’t the ones reading this thread.
In some fields all you need is a computer and an idea to be impactful, but in plenty of other fields you’d be hard pressed to make any credible, let alone meaningful impact without significant intellectual preparation and tacit knowledge. These things only come through experience, and for many people, the PhD program is that experience.
Disregard them. A lot of people fixate on the 1 in a billion celebrity exceptions like Musk, Thiel, Gates, Dyson, et al and go “look look you don’t need a PhD!”
Yes, a highly motivated college dropout with a computer, a strong financial safety net, and the right social connections can be in the right place at the right time to seize big opportunities. Most people are not in that position. Many high-impact technologies need more than what just a computer can do.
The main thing is to be self aware enough to know the path you’re on, what paths are available to you, and how to make the most of the connections and resources you have available to you. The second you start to get pigeonholed, wrap things up and move on.
Most discussions I see online about whether or not someone should do a PhD tend to assume:
- The student becomes hyper focused and pigeonholed into some esoteric and unemployable domain, destined to run on the postdoctoral treadmill for decades.
- The PI is a control freak who only cares about publications, and considers students who leave for industry jobs after graduation to be failures.
These stereotypes can have an element of truth, but there are more enlightened PhD programs and PIs that understand the value of cross-cutting and commercializable research than you’d expect from the discourse. Not everyone is stuck working on a pinprick of knowledge, and if you choose your program and PI wisely, you can go much further and do many more things than you would never have access to with just an undergraduate background.
This might be a good candidate for the 2025 Ig Nobel prize in physics [1].
More seriously, NV centers are one of the most accessible quantum hardware platforms, and do very sensitive measurements on all kinds of interesting stuff.
The idea of neurosymbolic systems has been in the air a long time, but every time I look at the commentary of an article like this I’m surprised at number the “OMG why didn’t anyone think of this?” type of comments.
For a while I got the impression that an ideological undercurrent of “DL vs GOFAI” had gotten in the way of more widespread exploration of these ideas. Tao’s writing here changed my view to something more pragmatic, that being the formalization of the symbolic part of neurosymbolic AI requires too much manual intervention to easily scale. He is likely onto something by having an LLM in the loop with another system like Lean or Athena to iterate on the formalization process.
There must be some kind of inside joke at CERN about this. I’ve sat through more than a few conference talks by CERN researchers (not even LHC affiliated) where Comic Sans was the default font on the presentation slides.
Fortran is alive and well in science and engineering. The more modern standards are much nicer to work with, but largely backwards compatible with stuff written 50 years ago.
More seriously, spin polarized D-T fusion is known to have an enhanced reaction cross section, so there are labs out there researching how to implement it more reliably.
One possible nitpick: were you aware of the naming collision with Versor, the general purpose GA library in templated C++?
http://versor.mat.ucsb.edu/