Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2013-11-02

| Channels | #pdl index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
04:13 _vicash_ left #pdl
10:05 drrho joined #pdl
12:32 drrho joined #pdl
21:53 pdurbin The Big Data Brain Drain: Why Science is in Trouble - http://jakevdp.github.io/blog/2013/10/26/big-data-brain-drain/
22:22 jberger_ pdurbin, I see your link and raise you : http://www.nytimes.com/2013/11/03/education/edlife/finding-life-after-academia-and-not-feeling-bad-about-it.html?_r=0
22:25 jberger_ pdurbin, did I ever show you how nice my scripts for my simulation look? The were meant for reusability
22:25 jberger_ https://github.com/jberger/Physics-UEMColumn/blob/master/examples/Cooke.pl
22:29 sivoais speaking of, this paper has some good high-level rules for reproducibility: "Ten Simple Rules for Reproducible Computational Research" <http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1003285>
22:31 jberger_ sivoais, you can almost completely reproduce my entire thesis from scripts etc in my thesis repository (other than some of the most raw data obviously)
22:31 * jberger_ pats himself on the back
22:32 sivoais :-D
22:32 jberger_ c.f. https://github.com/jberger/Thesis/tree/master/inc/model_results/rf_cav
22:32 Mithaldu huh, that first link explains nicely why many scientists i know are very bad developers
22:33 Mithaldu "While software-focused jobs do exist within academia, they tend to be lower-paid positions"
22:33 jberger_ I've seen this too
22:33 jberger_ actually its something I've been thinking about
22:33 Mithaldu yep, one time i was considering a job in bio-eng and didn't take it because they offered my basically nothing
22:33 jberger_ I've been wondering if I could be an academic computing consultant
22:34 Mithaldu telling me "you have no bio degree, so we can't pay you more"
22:34 jberger_ something someone could budget into their grant proposal
22:34 Mithaldu entirely ignoring the fact that i have a skill and experience set they desperately need and which i built myself in just as much time as their degrees
22:34 jberger_ "they desperately need" omg yes
22:35 Mithaldu i can say that with confidence because:
22:35 Mithaldu when i was at gene campus in cambridge and perused their library for perl and bioperl books the books were uniformly terrible
22:35 Mithaldu not just out-dated and perl 4 smelling
22:35 Mithaldu they taught outright dangerous techniques
22:36 jberger_ I knew someone who really just needed a data-persistence solution (JSON blobs would have worked fine) but they did it by copy-and-pasting huge chunks of output form Mathematica and then pasting thing into the input line in a new file!
22:36 jberger_ Mithaldu, eek, like what? goto?
22:36 sivoais some people are trying to address quality of science sw dev by teaching <http://software-carpentry.org/>
22:36 Mithaldu that too
22:36 Mithaldu but mostly stuff like
22:37 Mithaldu "here's how you build a web app, first you print, then you fetch a db row, then you print more, then you fetch another row"
22:37 Mithaldu or
22:37 Mithaldu "here's db example code, of course you put the password directly in your code"
22:37 Mithaldu also: bareword filehandles
22:37 * jberger_ facepalm
22:37 Mithaldu sivoais: i don't think teaching is the right way
22:38 jberger_ its funny this should come up now, I'm giving the keynote at YAPC::Brazil and this is essentially my topic
22:38 jberger_ http://2013.yapcbrasil.org.br/talk/15-1
22:38 Mithaldu you only end up reducing the time scientists spend on research while ending up with mediocre results
22:38 Mithaldu what you need to do is get great scientists workiing with, and more importantly, respecing as equals, great developers
22:39 jberger_ Mithaldu, that may be true, but I think that the incredible lack of programming education given to science students is apalling
22:39 jberger_ surely there can be some amount of "best practices" taught
22:39 Mithaldu jberger_: if you leave it to that they will end up with java :)
22:39 Mithaldu you *know* schools who teach programming love that shit
22:39 Mithaldu also
22:39 Mithaldu it's not about best practices
22:40 Mithaldu programming is a mindset of communication
22:40 Mithaldu you can't teach that with a few rules of thumb
22:40 jberger_ ok, I don't mean best practices, I mean any practices
22:40 Mithaldu just like you can't teach proper biology research methodology with a few best practices
22:40 jberger_ most grad students I worked with (myself included) taught ourselves programming out of necessity
22:40 Mithaldu that's what i mean with "respect as equals"
22:41 Mithaldu if they were to be taught one rule it's this:
22:41 jberger_ I was among the few geeky enough to try to learn it well
22:41 Mithaldu go to the community of the language and talk to them
22:41 jberger_ sivoais, cool link btw
22:41 Mithaldu that's another thing
22:41 jberger_ Mithaldu, preach!
22:41 Mithaldu scientists don't do "internet" in my experience
22:42 Mithaldu because the scientist groups at different schools are often in direct competition for grants
22:42 Mithaldu jberger: i have no scientists around me anymore
22:42 Mithaldu (except for you :D)
22:42 sivoais hehe, which is quite ironic given why things like NSFNET were created
22:43 Mithaldu granted: my experience was in europe
22:43 Mithaldu gene campus was working at a gene cross-referencing tool
22:43 Mithaldu and such was cirad in france
22:43 jberger_ "Those with the skills mentioned in this article could easily ask for several times that compensation in a first-year industry job"
22:44 Mithaldu and the cirad project eventually faltered
22:44 Mithaldu since they couldn't get gene campus to cooperate with them
22:44 Mithaldu and gene campus had more funding
22:44 jberger_ I got a job at a large bank writing boring software and I must admit, even if the grant budgets open up, I would to take a massive salary knock going back to anything like science
22:45 Mithaldu banks are a bit special though :)
22:45 jberger_ I'm not doing quant (and being paid what they are), I'm making a GUI trading application for internal use
22:46 jberger_ really I could be doing this almost anywhere and probably get paid similarly, except in academia
22:46 pdurbin whoa. I woke this channel up
22:47 jberger_ Mithaldu, I've seen similar problems even in my tiny field, there are a few groups who refuse to work together, and I'm sure we've all suffered, but such is life I think
22:47 jberger_ pdurbin, :-)
22:48 Mithaldu jberger: right, the thing is: it doesn't happen in software development
22:48 Mithaldu which is another reason for bad scientist developers, compared to just developers
22:49 Mithaldu they're not used to being able to cast a wide net
22:49 jberger_ true
22:49 jberger_ as I said, I would love to start a consultancy firm for researchers
22:50 Mithaldu err
22:50 Mithaldu for 20$ an hour? :P
22:50 jberger_ I know that many grants now require data warehousing, that could at least be the foot in the door
22:50 Mithaldu let me put it like this
22:51 Mithaldu i was probably the best perl developer gene campus has ever seen (judging by their code)
22:51 Mithaldu they offered me 28k gbp
22:53 jberger_ right, I don't think they would hire people of our caliber
22:53 jberger_ but if they make it a line item in their grant proposals for consultancy, I think there is a better chance
22:54 perigrin well and you get three of them and you're making $60 / hr
22:54 Mithaldu jberger: hrm, that sounds like politics, i know less about that :)
22:55 perigrin less politics and more "the money of science"
22:55 jberger_ if they propose several million dollar for a major instrumentation project, they can surely afford and could probably sell to funding agencies, the need for $30k of software consultancy
22:55 jberger_ we have a $50k table!
22:57 Mithaldu ... what does the table do?
22:58 jberger_ computer controlled active vibration isolation
22:58 jberger_ think noise-cancelling headphones, but rather than speakers, moving table legs!
22:58 Mithaldu can't tell if you just called that table a sex toy or are serious
22:58 Mithaldu wow
22:58 jberger_ hehe
22:59 jberger_ yeah, it is quite a thing
22:59 jberger_ we also have an $18k vacuum pump
22:59 jberger_ the turbine is magnetic levitation
22:59 jberger_ and the backing pump is a scroll pump
23:00 jberger_ translation: no part-on-part wear, no vibration, no oil
23:00 jberger_ what I mean to say is, if you are asking a funding agency to buy this kind of stuff, I think you can reasonably ask for software consultancy
23:01 jberger_ and as a line item next to those things I think it would look more than sane
23:01 Mithaldu hrm, yeah
23:02 Mithaldu well hey, if you ever get around to it, give me a poke, i've dealt with french scientists successfully, the only thing that i can imagine being worse would be french-canadian ones :D
23:03 sivoais here a business case: if you're automating all that expensive hardware, it really pays to do testing on the code :-) I'm always on edge when coding for special hardware because of the costs involved.
23:05 Mithaldu saw a post today by a dude working on radiation treatment hardware
23:06 jberger_ sivoais, exactly
23:06 Mithaldu they found out by accident that they've been running for years with production software that switched x/y coordinates
23:06 jberger_ hahahaha
23:06 Mithaldu luckily it worked out because nobody ever tried to apply it in such a way that that would surface
23:06 Mithaldu also
23:06 Mithaldu they had to use production hardware for testing
23:07 Mithaldu and they could only test between 1800 and 0600 and on weekends
23:07 Mithaldu so, uh, hope you never need that :D
23:07 jberger_ well that would probably be a common thing
23:08 jberger_ when you are building a prototype $3m piece of hardware, you can't possibly build a second one for testing
23:08 Mithaldu otoh
23:09 Mithaldu imagine the kind of state it puts developers in
23:09 Mithaldu developers responsible for software that can kill people
23:09 jberger_ Mithaldu, I agree with you, but I don't think that is likely to change
23:10 Mithaldu yeah, that's sadly likely the case

| Channels | #pdl index | Today | | Search | Google Search | Plain-Text | summary