It's like you're trying to exclude Cecily.
Well, maybe if you had put it in all caps.
Dropping the volume is brilliant. Wish I'd thought of that back when I had a classroom. Instead I would talk louder. In retrospect, that was counterproductive.
This is like Jack Donaghy's advice for dealing with a high-maintenance, manipulative mother! 1. Say no. 2. Speak low. 3. Let her go.
My advisor just described me as "a good talker". It was nice of him to say something nice after I presented my research in lab meeting, and we both knew "that was good research" wasn't in the cards.
"Mr. Tweety has a very nice bicycle. I've seen it!"
I think it could be good research at some point. I refrained from making the many excuses (I wasn't supposed to present for a month! I didn't get to start trying to analyze the data until Friday! I had to rewrite a whole bunch of backend code to even get access to the data! I haven't had the chance to talk about how to approach this data with anybody!) that strike me as perfectly valid, since that's generally the right approach.
OP: I'm reminded (perhaps dangerously and inappropriately so) of the following from Kahneman:
Half of them saw the puzzles in a small font in washed-out gray print. The puzzles were legible, but the font induced cognitive strain. The results tell a clear story: 90% of the students who saw the CRT in normal font made at least one mistake in the test, but the proportion dropped to 35% when the font was barely legible. You read this correctly: performance was better with the bad font.The puzzles were of the classic "easy unless you read it too fast and jump to a wrong conclusion" kind. But it is not clear to me whether more learning of the actual material goes on in the brows furrowed, working to catch each word mode than when everything is more relaxed. I can imagine arguments either way--the sustainability of the intensity by the listeners through the whole lecture is potentially an issue, even if it is better in the short-term.
9: I have always depended on the kindness of those in positions of authority. Admittedly to mixed results. But without knowing the details, it is generally their response to that kind of thing (being fundamentally on track, but maybe a bit unprepared or whatever versus truly floundering) which has separated my good bosses from the bad*.
*At least as judged from my perspective.
10: There are those similar Psych 101 studies where any change made to the environment of a factory - lights brighter! lights dimmer. Colder! Hotter. etc -increases productivity...for a couple days, and then it wears off.
I just got some good feedback (to wit: I had good stuff there but should have sold it more confidently and let other people pick at it if they wanted rather than pre-emptively anticipating criticisms) that I can work with. Spiral of self-doubt: provisionally averted, for now! Maybe!
I'm reminded of the following from Kahneman
Is good stuff, eh?
14: Yes, but only 1/4 through and just returned my one-week loaner since too much other shit going on. So am a bit frustrated at the moment, but will get time to finish it soon enough. I feel I need enough time to System 2 it.
14: I'm reading the book. It's very good. I was worrying that Kahneman was going to get Sunstein/Gladwellised, but no, it's a proper book.
It's the opposite of excluding me! I will be the ONLY ONE who can understand Heebie.
People always get this wrong about noisy bars too. No, I don't care how noisy it is. I'LL be fine. It's all y'all whiny "oooh poor me, if there's no sound input I can't understand English" suckers who have problems in these situations.
I do actually remember thinking that talking to you in a bar -- I don't remember if I said it, or figured out that you were having less trouble than I was before bringing it up.
Well, I think the issue is that it's hard for people to understand how little I can hear. Which I am sympathetic to, I just think it's funny. I already can't hear you! I'm not going to have any additional problems unless somebody turns off all the lights.*
*This is an actual problem in bars. I hate mood lighting very much.
Messily are there are studies of implicit lip-reading in people who have shitty hearing but aren't deaf?
I've long noticed that's what I do when I'm in a loud environment (look at the person's mouth) without it occuring to me to wonder if I've learned to be any better at it extracting meaning from mouth shape per se than somebody who hears well would be.
I guess I could look this up myself. But I don't feel like looking things up today because nyah stupid science.
I really like the phrase "Gricean cockpunch," but I'm worried that I'll never find another occasion to use it where the audience will appreciate its brilliance.
19: Yeah, I was also having trouble remembering you were deaf, given the ease with which you were conversing.
20: I don't know if I lipread better than someone with more perfect hearing might, but I know there's a big difference between what I can 'hear' if I can see the speaker's mouth and if I can't.
22.last: right.
I wonder if the McGurk effect would be stronger for hard-of-hearing people?
20: Yes, sort of. It looks like there aren't reliable differences. But I don't really know much about the area, or if there's any sort of consensus about what all is involved. (really that's more audiology/speech therapy than linguistics, at least at this school)
given the ease with which you were conversing
I'm not sure "ease" is exactly the right word...
(Actually, I do use this with disruptive classes - drop the volume of your voice, and let them respond accordingly.)
This also (sometimes) works on crying children.
I know there's a big difference between what I can 'hear' if I can see the speaker's mouth and if I can't.
Likewise, when watching a dubbed movie, I will "hear" almost half the dialogue I would otherwise. God, overdubbing annoys me so, so much.
20: I used to get into all sorts of Kafkaesque conversations with the DE and others because I would force muffled sounds into meaning something, often incorrectly. My lip reading, never consciously attempted, didn't compensate. Perhaps that's why we got along so well.
The hearing aid makes a major difference, even in noisy environments, the remote control is very handy for optimizing it.
I started reading the Kahneman book the other day too. It's fun!
I've also started learning R.
29.2: sweet!
I have been learning the hell out of doing data analysis in matlab, which is convincing the hell out of me that matlab is an annoying platform for data analysis.
Interesting. I was talking to people this afternoon who were arguing about the relative merits of R and Matlab.
I think matlab is actually a great platform. It just is too general-purpose for me when I'm trying to do stuff quickly while thinking about other things.
Almost certainly because I don't know it well enough.
Which is infuriating.
So, basically, don't listen to me.
I assume this means essear has taken up the cause of evil.
Oh also matlab chokes when trying to deal with giant data sets in the manner which is most efficient for matlab. Which, okay.
33: I'm dipping my toes in it, at least. And fomenting an insurrection among the youth. There's something vaguely guilt-inducing about having a discussion about how we all might abandon our field within five years when we're racking up a several hundred dollar restaurant bill paid for by our semi-benevolent bosses.
I had a conversation today with one of my fellow grad students about how I would appreciate it if he just talked a little bit less about how I wasn't particularly planning to find an academic job after my degree.
One of the grad students in my group has been simultaneously working on an MBA without her advisor's knowledge. It's pretty impressive subterfuge, really.
That's fantastic! I've just been setting up meetings with some startup-involved people, but I don't think anybody in my lab would be surprised I'm doing that.
37: Where does he keep his heart when he's off taking MBA classes?
The first rule of Unfogged grad school is don't reveal confidences to anybody in your grad school. Post them here instead.
34: I've found that too, especially on routines based off of Arpack.
35: I was actually convinced by whomever it was in that thread with actual experience (Spike? Cyrus?) that there was a genuine need, so I'm mostly joking and it sounds like fun. Still, you can't quit before explaining dark matter.
Another postdoc recently got a job offer from a local startup. I hope he sticks around another year; we have some fun papers to finish. But the frustrating thing for all of us is that Nature just isn't cooperating. Why you gotta be that way, Nature? Just throw us one unexpected signal to think about.
Just throw us one unexpected signal to think about.
Oh you should really consider the biological sciences.
I should say sorry, Spyrus, you're my personal Mr. Smablearcasings.
There's something vaguely guilt-inducing about having a discussion about how we all might abandon our field within five years when we're racking up a several hundred dollar restaurant bill paid for by our semi-benevolent bosses.
Hey, it beats the alternative.
I was briefly confused as to whether that was Nature, the ultimate reality; or Nature, the journal about Nature.
Well, in that case...
43: Why wait until retirement.
46, in not being 48, is impressive.
37: I am taking side classes elsewhere while in grad school, although the rate of enrollment and attendance has slowed to comical levels this year. Just tonight I tried to drive to class, got stuck in traffic, turned around and bought a bunch of junk food to make the trip worthwhile -- then home. Attendance is not strictly required, but it may be more required than what I've done.
How long is your colleague's MBA going to take? (Do the classes always meet at night to accommodate people's corporate jobs? I know nothing of this stuff.)
Almost certainly because I don't know it well enough.
not really - it's because matlab is basically a numerical computing program, not a statistics program. You can use it for data analysis because linear algebra is linear algebra and optimisation is optimisation, but you're unlikely to do a better job than the professionals who wrote STATA or the kind-of-professionals who wrote R. It's sort of like rolling your own cigarettes.
(the "rolling your own cigarettes" analogy first developed by me back in the days when some people still thought that there was any point at all to being able to write in GAUSS).
(Actually I quite like GAUSS. I *suppose* that if you had a huge amount of data that you needed to analyse veryveryveryfast all the time, and for some reason "buy a bigger computer" wasn't an option, GAUSS would be quite useful. But give it a rest Maynard, we're not bloody Deutsche Bank here)
37 is amazing. I do not quite understand how it is actually possible.
I do not quite understand how it is actually possible.
There are MBAs and MBAs. If Essear's student had been moonlighting at HBS, I'd be surprised, but there are "qualifications" out there that the average Unfoggeteer could phone in in a couple of weekends and not all employers understand the difference.
Most MBA courses, including some of the best ones, have part-time options. I did my masters in Finance while working a full time job, so I guess I could have done it while doing another degree.
56. Except that most of these are set up to accommodate students who are paid for by their employers, and the assumption is that the employer will arrange for them to have time to attend lectures once a week, etc. Doing a graduate degree in a hard science may be less flexible.
I would recommend anybody who's tempted by this to undertake the dsquared however-many-minutes-it added-up-to-in-the-end MBA.
Ha except that some of us *ahem* can't read that blog any more.
Except that most of these are set up to accommodate students who are paid for by their employers, and the assumption is that the employer will arrange for them to have time to attend lectures once a week, etc
If only. The part time program has lectures and classes from 7pm to 9pm and I got nothin'. You might be right about the MBA track having daytime classes though. The real thing that I suspect makes it possible is that the MBA type courses take it for granted that the students have other things to do in their lives which might be more important, so the requirements for attending classes etc are very lax - if you are a good private study, you can miss a lot of them. Also the projects are typically group work and it's explicitly part of the deal (it's management!) that the work groups will share out the assignments and cover for each other.
59. Keir, if you send Daniel your email, he'll probably let you in.
I could certainly have done another degree while I was working on my D.Phil, if I'd had the money. As it was I was in nominally full-time study and working a full-time job. Technically breaking the rules, but I got fucked over by funding rule changes and had no choice. At one point I was studying full-time, working full-time, and doing a half-time teaching post on the side for a somewhat mental total number of hours per week.
Which, in retrospect, basically meant doing somewhat shittily at all of those things, but if I'd had no financial pressures, and the two subjects were sufficiently different, I don't imagine it'd have been hard doing just two things at once.
I think once someone is enrolled in the university they're free to sign up for classes in any of the schools? I'm not sure how the actual degree-granting part works out. But I know someone else who turned a physics postdoc at Berkeley into a business school degree, several years ago. He became fairly unproductive on the physics end while getting the MBA, and then left to make heaps of money.
52: I don't disagree with what you're saying, but nonetheless in this case I'm pretty sure the biggest stumbling block is my incompetence.
You know what's great to be able to use for computation? C++. It is fast.
I got stuff done in seconds while MATLAB was still saying, "How about never - is never good for you?" (Well, minutes, but that might as well be neverl)
C++ sure has a lot of overhead. Why wouldn't you use assembly language?
C++. It is fast.
You know what isn't fast? Writing fancy numerical libraries in C++ to replicate the massive amounts of work someone else already put into making Matlab or Mathematica or whatever.
66, 67: For some problems, sure.
I am just excited that I have choices, and that they're sufficiently different to be real choices. I never realized how big the difference in speed was for some things, until I tried the same thing in both.
66: It's been a while since I coded in assembler. Does it let you chunk tasks/routines at all?
Also, isn't most of the C++ overhead a compile-time issue rather than runtime? If so, then it doesn't make a difference for a sufficiently large number of calculations. And if the task is small, then MATLAB is perfectly adequate.
Writing fancy numerical libraries in C++ to replicate the massive amounts of work someone else already put into making fancy, highly optimized numerical libraries written in C++ that form the bulk of Matlab's backend.
Which is to say, Matlab is extremely fast at lots of things, and incredibly efficient (if you're not me two days ago) at linear algebra-type operations, among other things. Just avoid for loops.
Yeah. I suppose if you find or write a user-friendly C++ wrapper for BLAS/LAPACK/etc you can do about as well.
Also, isn't most of the C++ overhead a compile-time issue rather than runtime? If so, then it doesn't make a difference for a sufficiently large number of calculations
give it a rest Maynard, we're not bloody Deutsche Bank here.
You know what's a really good numerical optimisation routine? "Solver" in Excel[1]. If your dataset is too big for Solver, then you really need to hand it over to the professionals.
[1] Oh yes it is.
But if you, like a younger essear once did, decide that writing a C++ library for matrix operations, SVD, etc is a good way to learn some math and programming, you are wasting your time. It'll take too long to write and end up being horribly slow and fail on all sorts of cases you didn't think to test. Go to a party or something instead, younger essear.
73: I wrote a crappy half-assed one for the educational value (learning how to write classes), and then switched over to a standard library.
72: I don't know how to use Excel, it's too complicated.
There are pictures. I don't understand how to use pictures.
Also my datasets are pretty much always too big for Excel
...laydeez