Thinking like a real <X>
-
The current fad in science education is all about getting the kids to "think like real scientists". Problem is, basically zero of the people making these curricula and doing the "research" have been anywhere near a real, working scientist. At most they're familiar with academic scientists, mainly tenured professors who are funded by grant money. From what I can tell, even just counting PhDs, only about 2-5% of all science PhDs end up as tenured (or tenure-track) professors. And some of those will be at teaching universities with little active research.
And even based on my experience as a graduate student, none of what they're actually pushing has more than a remote connection to the actual practice of science even at the academic level. Yes, you do a lot of reading and writing. But only a tiny tiny fraction of that is the kind that they practice with the kids. They're de-emphasizing content in favor of "skills". Because you can "look up content". Except they don't teach how to But the only way to understand the papers you're reading is to have the content on total lock-down. And the writing is very systematic and very different than the lab reports they're pushing.
I know based on discussions here (among other places) that academic computer science programs have very little to do with actual programming jobs (being way too theoretical, for one).
Am I just
here? Anyone here a practicing scientist (of any type, but doing research)? Care to weigh in on what the important parts of your job are?
-
@Benjamin-Hall said in Thinking like a real <X>:
I know based on discussions here (among other places) that academic computer science programs have very little to do with actual programming jobs (being way too theoretical, for one).
That actually varies very much (college-level work in CS ought to have a strong practical component because that reinforces the other learning), but there is indeed more theory than usual when at college typically because that's the key thing that many are missing. Done right, theory supercharges practice by both letting you understand what you're looking at more completely, and avoid a hell of a lot of dead ends. Most people don't do much theory after graduating, so it does need to be squeezed in.
Snark aside, one of the main things that junior scientists do is try to reproduce someone else's work. One of the most important features of science is that it is trying to build ways to predict the future, and scientific theories are formally only as good as the next test of them. The widely accepted theories are such because they've had a heck of a lot of testing! The usual first stages of this are learning to reproduce a widely accepted result (and in the process, getting the skills to do so), then trying to reproduce a newer result (i.e., only probably OK), and only then going on to really formulate their own stuff. That continues to be true for quite a bit after getting a postdoc science job (though the selection of what to work on changes, often being focused within their own lab).
You see less of that in CS, but that's mostly because CS is generally (comparatively!) easy to replicate the artifacts of and reproduce the results of. What you see more of is work on reproducing results on sometimes radically different systems, typically as a prelude to scaling things up/out. For example, we've been doing a lot of work lately on reproduction of simulation behaviours (of particular neuron models) that were originally computed using double precision floats, and getting that to work within a system that has only fixed point or software single precision floats and which is running in soft realtime. (This matters because the system can also scale up hugely, which just using Mathematica really can't.)
CS is at the intersection of Maths, Engineering and Psychology. (Absolutely everything to do with UI/UX is totally driven by psych.)
-
@dkf said in Thinking like a real <X>:
Absolutely everything to do with UI/UX is totally driven by psych.
Unfortunately, it often seems to be driven by psychos, or at least sadists.
-
I can see the value in changing the mindset of science education from "this is true because I, an authority figure, say so" to "lots of people have tried to work out why this happens and at this point we're basically certain it's because of that", and a key point of that can be "come up with a hypothesis, identify an experiment that can back up or disprove the hypothesis, do the experiment and see for yourself"
How similar or not that is to the actual day to day work of a scientist is moo IMO, since hardly any of the children in a given science class are going to go on to become professional scientists, but improving critical thinking and understanding of the scientific method is important in these days of fake news and heavy bias
-
@HardwareGeek said in Thinking like a real <X>:
@dkf said in Thinking like a real <X>:
Absolutely everything to do with UI/UX is totally driven by psych.
Unfortunately, it often seems to be driven by psychos, or at least sadists.
That's what you get when you have people doing it who pay no attention to psych.
-
Building on this, thinking about communication.
The primary ways I communicated as a graduate student were:
- Orally to a small expert group. This was me talking to my adviser or one of the other group members. Here, it's all about the content. Everything is as condensed as possible, with lots of unstated assumptions and relying heavily on shared knowledge. Lots of fast and furious back and forth. Definitely no concern for others' feelings at this point--the truth is all that matters. This was heavily multi-lateral--everyone is talking and adding things, asking questions. Basically free-flowing discussion that would be utterly incomprehensible to anyone outside. Often not even in complete sentences, with lots of thinking out loud.
- Orally to a large, expert-adjacent group. Poster presentations, talks, or thesis defense. These had visual aids in the form of a poster or a powerpoint presentation. These frequently are uninterrupted, with questions at the end. Clarity and conciseness are important here, and a steady mind to handle frequently hostile-framed questioning. The audience is not expert in the particular sub-field, but is expert in the slightly-larger parent field. Often this means that the questions and comments are framed in terms that differ slightly from your "home field" and you have to know/handle the differences in basic assumptions. Thinking on your feet is critical here. This is dominantly one-way transmission of information, with a little back-and-forth at the end.
- Written reports. Publishable papers, mainly. Here it's all about following conventions. The journal has conventions, the field has conventions, etc. Most of what English teachers teach about the "right way to write" is totally irrelevant or actively bad here--especially when it comes to repetition. There's no expectation that a scientific paper will be read straight through. Usually, the abstract (which no one teaches you to write, btw) is the most important, then the introduction, then the conclusion/results. Method sections are important for the few who settle down to try to reproduce or fully understand, but most don't even get that far. Passive voice is everywhere. Citations are essential, but use the journal's house style, which is rarely (in Physics at least) even remotely similar to APA or MLA. Totally one-way communication.
So learning to communicate (both output and input) is vital. Learning the content is critical--no, you can't verify most of it for yourself or develop even the most basic ideas on your own--that's thousands of man-years of the best minds. Standing on the shoulders of giants is the fundamental rule for modern science. Yes that involves heavy memorization as a foundation for understanding and application. If you have to think about something like the conservation laws or the basic methods for solving differential equations, you're screwed. You'll never get anything done, and you'll make mistakes. Those have to be automatic.
On the other hand, "experimental design" is something that the people writing the grants/managing the labs deal with. Most scientists, in my experience, are in the trenches. They're using well-worn techniques and applying them to slightly different substrates each time. Mucking about trying to upend the established experimental techniques is reserved for the tiny fraction of cutting-edge people. And that's something that can't really be taught well, because it's an art form more than anything. You have to have a feel for it, mainly based on long experience with the established methods.
So I'd want kids to mainly learn the following:
- Reading scientific writing and restating it in their own words.
- Writing scientifically. This may be taking "normal" work and writing it in the appropriate style, or taking regular "cookbook" labs and writing up the results properly.
- Talking/discussing. The small-expert-group stuff is hard, because none of them are experts and they're not comfortable with each other enough to go no-holds-barred. But the large-group-oral presentations might work.
-
@Benjamin-Hall said in Thinking like a real <X>:
Mucking about trying to upend the established experimental techniques is reserved for the tiny fraction of cutting-edge people.
It's called “method development” and it's rare as hen's teeth.
-
@Benjamin-Hall said in Thinking like a real <X>:
Anyone here a practicing scientist (of any type, but doing research)? Care to weigh in on what the important parts of your job are?
Not me, but there's a ton of interesting stuff here:
https://blogs.sciencemag.org/pipeline/
I'd suggest browsing the tags. This one in particular looks promising:
https://blogs.sciencemag.org/pipeline/archives/category/academia-vs-industry
-
@boomzilla That's actually one of the sources that made me think about this topic more. Especially since I was only on the academic side of the fence.
-
@dkf said in Thinking like a real <X>:
@HardwareGeek said in Thinking like a real <X>:
@dkf said in Thinking like a real <X>:
Absolutely everything to do with UI/UX is totally driven by psych.
Unfortunately, it often seems to be driven by psychos, or at least sadists.
That's what you get when you have people doing it who pay no attention to psych.
You're entitled to your opinion (yes, I know, that's big of me), but I think it would have no detrimental effect on UI's if psychologists were never consulted. (FWIW, my bachelor's was in Social and Behavioral Sciences.)
-
@jinpa said in Thinking like a real <X>:
no detrimental effect on UI's if psychologists were never consulted
In many cases that is correct... they are so bad that making them worse would be quite hard.
