Sign in or Join FriendFeed
FriendFeed is the easiest way to share online. Learn more »
Jen Dodd
Greg Wilson: Can the Web Make Scientists Brush Their Teeth?
If mom can't, what chance does the web have - Deepak Singh
We don't know much about how scientists use computers. - Jen Dodd
Those who talk about it tend to be early adopter or heavy users. - Jen Dodd
Real survey coming, but not ready yet; nor URL so as not to pollute data. - Chad Orzel
A small minority develop "big codes for big iron". - Jen Dodd
The majority develop for the desktop, and they don't go to scientific computing conferences. - Jen Dodd
Claim: LHC analysis code is going to produce nothing better than random numbers. - Jen Dodd
LHC analysis code would get failing grade in CS; Wilson thinks it will be basically random numbers - Chad Orzel
... as an example of a general issue with reliability and reproducibility. - Jen Dodd
Scientists don't care about stability, quality of code, as demonstrated by where we spend time. It's a fair cop. - Chad Orzel
Microsoft report on science software: - Chad Orzel
Excessive faith in heroic effort-- can meet any deadline by pushing harder - Chad Orzel
Ford studied the effect of length of the work week on performance of knowledge workers; concluded the 40 hour week was optimal. - Jen Dodd
IBM stretch imgae: -- love the suits. - Chad Orzel
So why shouldn't scientists spend more time working with software engineers. The model works - Deepak Singh
And why do scientists not respect good software engineering? - Deepak Singh
@Deepak - do you have an example of this? - Jen Dodd
Faintly insulting analogy between scientific computing and Third World economies - Chad Orzel
Claim that high performance computers are not well used. - Jen Dodd
Jen, first startup I worked at. Scientific programmers developed algo's, software engineers implemented algos. Rosetta. Other groups where software engineers are specifically hired to implement scientific code into robust platforms. Not common, but it happens and it works well. Of course, life would be easier if scientists wrote good code - Deepak Singh
That sounds like a sensible division of labour, since it's clearly a full-time job to become either a good scientist or a good software engineer. Maybe this is what's underlying Greg's observations. - Jen Dodd
Funny student exam answers - Sabine Hossenfelder
Provenance and workflow systems. See eg. - Jen Dodd
I don't really disagree with him, but I don't think this problem is unique to computing. What passes for circuit design, plumbing, machining, etc. in much of experimental physics would be just as appalling to professionals in those areas. - Chad Orzel
Greg: computer scientists should make tools that eliminate the need for scientists to "do the plumbing". - Jen Dodd
provenance challenge wiki: - Chad Orzel
betting on Yahoo pipes because their market is bigger - Paul Guinnessy
Jen ... what kind of tools? Software engineering has never been more interesting and had more interesting platforms - Deepak Singh
Deepak, not sure I can say for sure - did others understand that? - Jen Dodd
Pipeline Pilot, Taverna, Rails, Django, Hadoop, etc etc etc? There's so much stuff out there. People just don't use it. Agreed, good scientific frameworks are not quite there, but there is a starting point. - Deepak Singh
Buckheit Reproducible Research article: - Chad Orzel
@Deepak, that's probably what he was talking about - he says he thinks that existing commercial products will develop to fill this need. - Jen Dodd
I know a few software engineers who've left academic groups because no one cared about good quality code, they just wanted things to work. - Deepak Singh
reproducible research link: - Chad Orzel
Jen, they've been there for a while. There are accessibility issues and programming frameworks are fairly mature. The level of computer science and higher level mathematics being used at many web companies is very advanced and they do a good job (a very good job in fact). It's an attitude and approach problem. Neil's points above are quite valid - Deepak Singh
Using the word "knuth" as noun to mean a paper plus the code in a readable format (?) - Jen Dodd
Advocating "literate programming" for scientists - Chad Orzel
Deepak and Neil, I can't speak from personal knowledge, so it's interesting to hear what you think about Greg's perspective. - Jen Dodd
Greg: Agile development practices fit scientists's needs better than traditional approaches. - Jen Dodd
Software engineering as "the youngest science". - Jen Dodd
One-term course to boost productivity 20%: - Chad Orzel
Software carpentry site will be relaunched as a wiki. Difficulty of institutional barriers to offering courses that cross areas such as this. - carolh
Lengthy political discussion, whose start I missed while dealing with funding issue in email. - Chad Orzel
most people that you talk to over the web are angry. sooner or later any political action stops being polite. says that ultranationalism and facism in Russia is getting worse, “Oil is to the 21st Century what slavery was to the 19th: morally repugnant, but economically indispensable” Hochschild said an interview, Bury the chains “Adam Hochschild” good book on slavery -- they (quakers) invented a lot of the terms we use in democracy today. has a higher ranking than - Paul Guinnessy
can scientists prove that they didn't fake their data? Lawyers may want to know. - Paul Guinnessy
Agile works well in certain situations. Matt Wood can comment extensively on that. But there needs to be some recognition of the importance of good software within an academic setting - Deepak Singh