Conversation Between SunderX and DYKWIA

They're important to different degrees in different jobs. For example, a cubicle programming gig is going to need less social competence than an IT consultant.

A university draws a line somewhere between objective education and `life training'. A university should, at the very least, make someone more knowledgeable about a certain trade. But where does this stop? Should a university also socially condition students into what they think their preferred model is? Should they lecture morals as well as facts? Should they espouse a certain righteous political viewpoint, and craft their vision of `leaders of tomorrow'?

My experience as a computer scientist has shown me that people are vastly different and everyone finds their own niche. Maybe it's different elsewhere, but in Cambridge it's okay to be a geek and fascinated in something, and to have to disappear for a couple of days to do work. Forcing introverts into social situations seems counterproductive, as those that want to develop such skills can certainly try and are presented with plenty of opportunity. Those that don't are given the opportunity to go their own way and develop themselves as they want - whether this leads to PhD research, working in a quant, or as a cubicle programmer.

The way that American universities select applicants, on the basis of starry-eyed, exaggerated admissions essays, indicates a preference precisely for the above type of social selection. It's not necessarily wrong, but I don't agree with it.

Teamwork and social skills are essential in most jobs, just as much as being knowledgeable. IMO it is one of the things a college should be teaching. If you hadn't noticed a lot of CS students are socially awkward and probably wont change on their own. Also, where did you get the idea that only outgoing students get into college?

Phew. So I think that the lack of hands-on work in Cambridge is grossly exaggerated. Clearly based in fact, but blown out of proportion. ISO may talk all she likes about Oxbridge, but afaik she's not there, and therefore can't reliably comment.

Teamwork and social skills are clearly important in the workplace, but the development of those is up to the individual and not the university. There's plenty of opportunity to join societies and committees in Oxbridge, and given the collegiate system there's more opportunity for local student government and socs than anywhere else. I very much agree with the UK style of ignoring most extra-curricular activities and focusing on academic profiles. You go to university to learn, not to tap-dance for underprivileged Nigerian children. The socially awkward deserve a chance at education as well.

Well, it's simply not true that they refuse to teach vocational things. It's just that there's less time overall to give to students to muck about. Consider that terms in Cambridge are 8 weeks long, and a first year computer scientist will be studying not only CS but typically Physics and Mathematics at the same time.

There's a significant dissertation/project work component in third year and a couple of group projects before that. Admittedly, pretty much all projects in university are designed to be time sinks, but it's a good opportunity to do your own thing for a while.

As an example of the `vocational' stuff I did as an undergrad:

* I implemented Pong in hardware on an FPGA board with VGA output
* Did the same with the Game of Life, but on a simulated MIPS core, programmed in Assembly
* Produced a routing app for the lab based upon image recognition of rooms
* Countless numbers of Physics labs, including building a radio receiver, etc
* A full-motion gesture recognition application for presentation control using motion-capture bulbs, Vicon cameras and Hidden Markov Models.

And in the MPhil:

* Designed my own spam classifier, using a new approach involving Latent Dirichlet Allocation (a form of topic modelling), and SVM libraries.
* Implemented part of a speech recognition engine, translating from waveform data into transcripted phonemes, trained on several hours of broadcast TV audio.
* Designed and programmed an adaptable VDU out of a gazillion LEDs and what was essentially a washing machine controller, and programmed it to run Snake with a keypad.
* Implemented a gesture recognition app for Android phones, using 3D accelerometer data to track distinctive movements.
* Built a new music recommender using data from 360k users from Last.fm, incorporating 3 new algorithms that I proposed. Being published as a paper in WSDM 2012.

nothing in particular. ISO has a particular obsession with ox/cam, which is perhaps slightly unfounded. Again, I wouldn't say CS at cambridge is less than one of the best in the world, I just feel it is a very different approach to teaching. I'm not so sure it leads to better grads though. From what I have heard there is less focus on teamwork and social skills - which might lead to 'better', more knowledgeable graduates but are they more employable (esp. outside of research fields)? Also, I'm guessing faculty there have a slight aversion to teaching anything remotely vocational - which probably leads to the more theoretical approach.

Well, Cambridge CS is pretty much the top of the UK, and probably Europe as well. Having been in postgraduate research too, the staff and students at the Computer Lab are top-class and their research output easily matches US universities with departments of a similar size.

The thing about Cambridge undergrad education is that yes, it is (potentially, depending on course choices) much more theoretical. I get the impression that US universities have a much more hands on approach to teaching and project work. But the question is who produces better students in the end. Pound for dollar, Cambridge manages to produce top class graduates at a much cheaper rate than US universities with massive endowments and hilarious student fees. Consider that it does so also with only 60-70 students in a class, in one fewer year, and in a country where CS itself is less culturally accepted.

I can't speak in particular about parallel computing, but there's certainly massive amounts of work going on in terms of multicore programming, hardware design and hadoop-style parallelisation. One of my friends is doing his PhD on a project which is going to simulate the neural interaction of a human brain with hundreds of FPGA's networked in parallel. Obviously different departments will have strengths in different areas, but what exactly are you suggesting about Cambridge research?

Oh, you graduated there. I wouldn't say Cambridge is especially bad. I think it has a very different approach to CS from the other top 4 CS colleges in the US. US colleges have more money, are more hands on and more application based - but I can see the benefit of a stronger mathematical foundation. Some just feel cambridge is slightly behind is all (ahem parallel computing ahem).