As a SW engineer I’m just amazed at how such a large scale physical engineering that’s on the cutting edge of science takes place. Truly some of the brightest minds on Earth must be working on this project to make sure the experiment itself is done properly and developing any new engineering science along the way to build new capabilities. Not to mention ensuring that there’s no fatal engineering mistakes along the way putting theory to practice.
Oh i see what you did there and where you come from.
Its actually how inbelieve Elon got successful .. its his physics training.
Ive met physics students in higher semesters ... and they were down right better programmers too (thry built things they used, they really hacked programs together) while we spent our time trying to ... well i dont know what, but somethig wasnt right. This wasnt CS50 though .. just a German uni.
The undergraduate portion of physics is the most general and therefore the most helpful for what Elon does (really broad scope systems engineering with an ability to go fractal deep into any technical detail). And I actually do think the physics degree was majorly helpful in the approach Elon takes to solving problems in new fields. Elon is smart, sure, but not the smartest person ever. You have to credit the physics approach for much of that (and a combination of tough life growing up and extended family that provided a safety net that enabled risk-taking... the latter part could be scaled but the former shouldn’t—and may have contributed to some of Elon’s flaws). The good thing is that undergraduate physics education can be scaled to a lot of people (and aspects of its approach to solving problems adopted by other fields).
I have a physics degree. I think an interesting thing about physics education is that it teaches resourcefulness and opportunism. Physics "owns" relatively few techniques and technologies. It borrows from others, and combines things.
One of the reasons physicists like programming so much, is that it's the modern duct tape for putting disparate things together, and there's so much stuff out there to use.
Physicists are always among the earliest users of technologies -- vacuum tubes, transistors, and successive generations of computers. At my college, the earliest adopters of personal computers were all physics professors. The one exception was a humanities prof whose kid happened to be a physics major.
In my industry (computational engineering) I fear leetcode etc. is becoming the bar one must pass before being questioned about anything germane (e.g. what kind of time integrator to use in this situation, different ways to handle that boundary condition, derive a numerical system for some physics with some spatial accuracy, etc)
This takes time away from me actually being the best numericist I can be, and puts me off doing leet-code challenges. It’s absolutely infuriating. When I arrived on the scene ~8 years ago we were isolated from that crap. No more. (It’s crap because there are better ways to test if we are competent (ie give me a multiresolution analysis problem on the proverbial white board, not a string puzzle, otherwise you throw the employee baby out with the bath water or make the baby memorize pointless (for this kind of baby) trivia.
I wonder if this is somehow a scheme to reduce employee turnover in general - to get salaries under control. I doubt it could so be planned. I have no doubt it has some effect in that direction as it adds friction and to the job hunters process.
I find it interesting/satisfying that clock tick is one big bold configuration parameter that can be tweaked to basically change the interpretation of the results.
It has been standard for a long time in particle physics. The experiments that I worked on were blinded in similar ways, though a bit less elaborately. On my would-be thesis experiment all the blinding turned out to be pointless... the signal was as bright as day. Always good when that happens!
I think this is possible because of the exclusivity of the equipment. These scientists have the luxury of being the only ones able to conduct this experiment, and presumably some long run funding to do it. If there were 5 different teams competing, there would be less rigour I suspect, since other metrics (being first, number of papers, the usual academic swill) would come into play.
I've been working with an econometrician, what really impresses me is her insistence that we don't do any analyses until we've figured out what the right ones to do are. No exploring because it will taint your p values! It's a hard discipline to follow.
Makes sense really. From ML; the test data are an unbiased estimate of the performance of a model and you can use it get PAC (probably approximately correct) estimations of your performance. There are three (4) powers at play here, the confidence [1], the error bound [2], and the number of data points (and model complexity [3] but that is not relevant when we compare a single model against test data). Your performance is exponential in the number of samples you have (meaning more data gives you exponentially better bounds and more confidence), alternatively, for a fixed sample, you can have either high confidence, but weak bound, or small bound but low confidence. When we consider the training data as an estimate of the expected performance, we also take into consideration the class complexity.
The moment you touch the test data you have tainted it because your brain can do a crude search on the space of solutions and reconcile them after the fact. This is why any preprocessing and cleaning is parameterized on the training set and then applied to the data, that also includes the class of models you can consider and is why you never select final models based on their performance on the test set; it becomes biased.
[1] probability of getting a good model.
[2] how much the empirical performance underestimates the expected performance.
[3] the number of models you can select from a particular class of models.
I am not qualified in any way to answer that question, but my amateur opinion is that as long as we are still so tribal as to have a number of different governments all serving different priorities, we'll never see humanity advance beyond our abilities to destroy each other and ourselves.
Considering that we didn't have a real conception of what physics was until Newton and didn't really get the ball rolling until Maxwell, we have moved at a pretty good pace. It's entirely possible that the fundamentals of physics are off limits until we become Kardashev Type II
> A notable early use of a blind analysis in physics was in a measurement of the e/m of the electron, by Dunnington. In this measurement, the e/m was proportional to the angle between the electron source and the detector. Dunnington asked his machinist to arbitrarily choose an angle around 340°. Only when the analysis was complete, and Dunnington was ready to publish a result, did he accurately measurethe hidden angle.
25 comments
[ 4.8 ms ] story [ 59.5 ms ] threadThey won't be able to invert a binary tree on white-board, or a hacker-rank tab, in 30-45 minutes.
They're clearly not the smartest ones.
Its actually how inbelieve Elon got successful .. its his physics training.
Ive met physics students in higher semesters ... and they were down right better programmers too (thry built things they used, they really hacked programs together) while we spent our time trying to ... well i dont know what, but somethig wasnt right. This wasnt CS50 though .. just a German uni.
One of the reasons physicists like programming so much, is that it's the modern duct tape for putting disparate things together, and there's so much stuff out there to use.
Physicists are always among the earliest users of technologies -- vacuum tubes, transistors, and successive generations of computers. At my college, the earliest adopters of personal computers were all physics professors. The one exception was a humanities prof whose kid happened to be a physics major.
I wonder if this is somehow a scheme to reduce employee turnover in general - to get salaries under control. I doubt it could so be planned. I have no doubt it has some effect in that direction as it adds friction and to the job hunters process.
Unluckily for them their online literacy says I'm too stupid to write simulation code them so their loss.
But imagine the FPGA-grunt in those things!
The moment you touch the test data you have tainted it because your brain can do a crude search on the space of solutions and reconcile them after the fact. This is why any preprocessing and cleaning is parameterized on the training set and then applied to the data, that also includes the class of models you can consider and is why you never select final models based on their performance on the test set; it becomes biased.
[1] probability of getting a good model.
[2] how much the empirical performance underestimates the expected performance.
[3] the number of models you can select from a particular class of models.
https://muon-gm2-theory.illinois.edu/white-paper/
> A notable early use of a blind analysis in physics was in a measurement of the e/m of the electron, by Dunnington. In this measurement, the e/m was proportional to the angle between the electron source and the detector. Dunnington asked his machinist to arbitrarily choose an angle around 340°. Only when the analysis was complete, and Dunnington was ready to publish a result, did he accurately measurethe hidden angle.
[1] https://www.slac.stanford.edu/econf/C030908/papers/TUIT001.p...