France excels at R&D, but Academia is overdue for disruption

I feel one big thing that the "cool" (as opposed to us 'uncool' academics) can miss is that good research is hard. It is common to spend months on topics on areas that don't work out, and years on topics that do. It doesn't fit the modern high-speed of the internet, and it doesn't fit in 140 characters.

And of course, most research isn't particularly interesting to people outside a fairly small area. I might come up with a new way of doing something important in my research area 50 times faster than we did it before, but explaining what I did struggles to fit in 15 pages, never mind 140 characters. Personally I can't imagine what interesting research related-comment you could fit in a tweet. Then again, I've never got twitter (which isn't something I'm proud of -- sufficiently many people are using it that I am sure I'm missing something.. just not sure what).

As someone tangentially involved in the administration apparatus for large, well utilized clusters, I have to dispel some myths related to "Cloud" services and Amazon EC2 for computation. First, if your school has institutional cluster services, these are FAR cheaper than cloud options. Secondly, the type of systems that are good for scientific computation are not the type of boxes typically deployed in the cloud; try to find a 48 terabyte memory node in the cloud.

OSG/Teragrid are much better options than the cloud, though you do have to jump through some hoops to access them. Believe me when I say it is worth it though.

The argument about France's strength at the end of the article is anecdotal (and not argued at all), and the title is misleading.

(and as a French CS student who did his undergrad in France and started a PhD in the US (and have many friends in PhD tracks in both systems), I'd actually argue that France is better at "pure/abstract research" rather than R&D. Inria is a very unusual organization, not representative of research in France at all)

I think this post misses a number of things. For example, take peer review. The system is not perfect--but peer review takes time and expertise. It's not as simple as a "like". When I review an article--I check for consistency. If necessary, I go through the references. At times, I make sure that errors have been propagated correctly. Then, there are broader issues such as relevance. In short, it's labor intensive and sometimes there are bad referees--but overall, I think the system is better than a simple comment system. In modern terms, you might call it "curated" content--the referees and editors help increase the signal to noise ratio in the sea of papers that are submitted.

For conferences--while the presentations are important, so are the conversations in the halls, at dinner, etc. It's useful to have face to face conversations with your collaborators and to make new ones.

There have been some changes--like for initial program committee meetings for conferences sometimes take place using webex/gotomeeting given the diverse timezones/countries of members. Some experimentation has been done with using second life for the first stage of grant reviews--but at the end, physical meetings are necessary. For conferences themselves (in physics), you are paying on average less than $1000--even for conferences with ~10,000 people. But--conferences do cost money to organize--I made a mistake for the first workshop I organized by charging too little...

Things are gradually improving--slides from the invited talks are hosted online for the major conference from my field. Mobile schedules could be improved...

Others have commented on the EC2 question--but one other thing to consider is whether their architecture is suited to a given problem. We're investigating this now.

While Hadoop is good for some problems, it may not be appropriate for others. This is generally true of a number of the points in the article--use the right tool for the right job--don't just use it because it is new...

There actually is a really strong community of mathematicians at mathoverflow.net. Also, I know quite a few academics who tweet about their research but these are definitely the exception rather than the norm.

A simple reason as to why academics aren't "socially connected" is that they don't need to be. The academic social network exists informally and is more tightly knit than a social network of say programmers, so I don't necessarily have to be your facebook friend or follow you on twitter to be able to get in touch with you through 2-3 hops

As far as blogs are concerned, one reason to write blogs is for purely didactic purposes. I definitely know tons of Grad students who use to prepare their notes and put them on their web pages or local servers to fulfill that need. The other reason: To transmit your ideas to wider audience is much a more harder problem to solve. Mathematical abstractions get progressively harder to communicate if your understanding of the basics is weak, example, you would definitely need a background in vector spaces to understand functional analysis. I remember reading a blog recently where the author ( a machine learning academic) had written about his experience in implementing a stochastic gradient solver for online learning in Clojure: There were a couple of comments complaining that there was no context. While, I definitely see where those folks are coming from, it would take too long and a good deal more posts to be able to break down a couple of book's worth of ideas before you get to what you want to talk about.

Reading and writing tweets is a pleasant time-waster that should be seriously limited if you want to achieve something serious, not encouraged. The same applies also to HN, of course, where I am now wasting time to avoid research :P

There are enough (and too much) scientific stuff to read in any interesting, active research topic; the bottleneck is in the researcher's ability to read and understand, not in lack of content.

And even peer reviewed articles in respectable publications quite often are garbage and/or wrong. If your proposal "let's do stuff this way" isn't even peer reviewed, then it's most likely not worth reading - since there's a too high chance that it's one of:

a) duplicate of something that was already done in 1970'ies by someone else, and abandoned as it's not practical in real life due to X,Y and Z that you haven't considered yet; b) misleading or simply not true, as it's not researched / repeated / analyzed enough; c) Simply useless.

After all, if it was correct, novel and interesting, then there are quite a lot incentives for putting it to 'traditional' publishing; so if you're not publishing it 'properly' then it's an indication that you are doing it because your content is not publish-worthy, as people tend to squeeze as much publications as they can out of any decent, publish-worthy research.

Actually, call me a twitter-blase, but I'm not so excited and upbeat about Twitter anymore. The fact that people tweet stuff during a presentation don't get me excited. I'd rather they listen and learn from the presenter. Then I can say mission accomplished (as a conference organizer).

I just know too many great people who don't spend any time on Twitter, and frankly, why should they? I see a lot of self-promotion, "social media" experts retweeting to death. Real people saying interesting things? Some, but not enough to keep me engaged.

I dread the day that scientists try to condense their research into 140 characters. Reading well thought out blog posts would be great, but to be honest, the ideal would simply making it easier for me to find out what the best papers are.

I always assumed that scientists generally do not participate in the wider web as much due to having much faster academic links than the average net user could even dream of.

However, that said, there are many things that can be improved upon that the wider web would be great for. Academic publishing is one obvious one and the Elsevier boycott is a strong indication that this is well underway.

Data collection and sensing is another, which with the new class of cheap networked devices like the Rasperry Pi, as well as pads and smartphones, could well be something that is worth putting some thought into.

Distributed computing and mass collaboration is another obvious one, however it seems that academics can often get this more or less for free, if you look at examples like folding@home and the NASA moon crater identification project thingy, so there might not be so much requirement for things like EC2 or mechanical turk, when compared to getting people to donate resources for nothing.