key bindings in kde4

July 22nd, 2008

I like software to be configurable and not designed for idiots. I want to be able to shoot myself in the foot.  I was put off gnome when they went the simpler-is-better (I-know-better-than-you) approach and the new kde seems to be taking a similar road.

A lot of key bindings are preassigned (not bad) and can not be changed (bad) because to do so would clash with some git’s preconceived notion of how to use a computer. Take assigning Ctrl-arrow to move around desktops which I’ve been using since before kde. Try to assign this in kde4 and you get an error:

The ‘Ctrl+Right’ key combination has already been allocated to the standard action “ForwardWord” that many applications use. You cannot use it for global shortcuts for this reason.

WTF? I switch desktops all the time, I don’t think I’ve ever felt the need to move the cursor forward a word.

Good software should be easy to understand and learn, but it should also be flexible - something that is often forgotten.

Luckily, it’s linux. Edit ~/.kde4/share/config/kglobalshortcutsrc to your heart’s desire.

Physics by pattern matching

June 26th, 2008

Yet more surprises marking exams … a question on friction … a number of answers began with

F = μ N

followed on the next line with

N = 6.02 x 1023

What?! This one threw me for a bit. After some searching I realised that Avogadro’s number was listed with other constants on the front of the exam. But, N was never used for the normal force in the notes or the textbook …. more searching found it in the formula sheet at the back of the exam.

So the question was approached as a pattern recognition exercise with absolutely no context, the access key was probably the symbol μ tiying it to friction…

Why is the Turing test supposed to be hard?

Exam questions going bad…

June 26th, 2008

I’m in the middle of marking exams for a course and have had a few suprises.

It’s pretty common for students to interpret or tackle a question in a way you hadn’t foreseen, and this is fine, you err on the side of the student. What I’ve just realised is that you shouldn’t offer life tips as part of the question as the context can got spectacularly awry:

I had a question examining a beam from a 1mW laser pointer entering the eye for the duration of a blink. Things like how much energy was transfered etc, then finishing with how the intensity compared against the image of the sun on a summers day. I finished the question with “NB though this value is high it does not appear to cause lasting damage for exposure within the blink time (0.25s), so laser pointers are usually classed as safe”  (they’re a bit touchy about laser pointers here in NSW).

What I hadn’t counted on was some students interpreting mW as a mega-watt and therefore tacitly assuring them they could look into a megawatt laser … hmm, maybe I should have said it would not cause lasting damage, again.

Citation trading

June 19th, 2008

One of the main currencies for academics is citations. For better or worse, reputation, funding, jobs and promotions ride on publications and citations. It’s popular to use citation based metrics such as the h-index to rank people. Highly cited papers increase a journals impact factor, making the journal desirable to publish in, and in turn, publishing in high-impact journals increases the perceived importance of the work and researcher …

h-index

The h-index itself is an interesting metric of performance that’s all the rage with job-selection and promotion committees at the moment. Essentially if you list someones publications ranked in decreasing order of citations, the h-index is the size of the largest square you could draw under that graph. Why a square? Who cares, arbitrary metrics answer to no one. It’s popular. Deal with it. The interesting point is that there’s often only a couple of papers that define the top right corner of the square and a couple of extra citations there will bump up the h-index.

The trouble is, despite their influence, citations are a pretty crude measure of how important a piece of work is. Any context to the citation is lost, all citations contribute equally (you can’t anti-cite a work, or add weight to a given citation - a citation from a grad student is as good as one from a Nobel laureate). In fact, for many researchers doing the citations is seen as a chore rather than a bestowing of praise, and there is a tendency to just copy the introduction to a field and all its citations from a previous work (research[1] shows citation errors tend to propagate). The take home message is that you have an enormous amount of flexibility in what you cite.

So, here is a wonderful evil idea. Why not set up a citation trading system? All you need is a website and access to citation data.

Here is the simplest way it could work. Every time you get cited by someone you owe them a citation. You can pay them back by citing some of their work - remember you have a lot of flexibility in what you cite. To make it interesting, citing particular papers can be worth more or less depending on whether they help bump up the h-index of an author.

These citation balances can be calculated automatically and made available on a website which can be viewed anonymously, so no one need know you’re playing the system. The citation worth of a paper is also an automatic calculation from citation data.

Researchers will develop different credit ratings (how well they repay their debts). Someone with a great credit rating is a great person to cite as they are much more likely to cite your work in return.

Hell, even just having an up-to-date list of the boundary papers for someones h-index would be great - here a single citation could land them a promotion: what a nice way of giving someone a pat on the back.

[1] If you don’t have access to my high-impact souce take a look on the arxiv

Free textbooks

May 22nd, 2008

I recently acquired four physics textbooks and was shocked to discover that they cover exactly the same material in exactly the same way. Sure the diagrams are different and the problems are different, but not that different. Looking for something more original I started searching for stuff online…

Here’s two sites I’ve found that have collections of free textbooks:

powerpoint 2007 - ink sucks

May 11th, 2008

Well, tried to use powerpoint 2007 to prepare some lectures today only to find a nasty surprise … ink support has taken a giant step backwards and is just about unusable. It seems that the intrepid programmers at Microsoft are colossally shortsighted and only see ink as useful for temporary annotations: you cannot select the ink by dragging a selection rectangle or using select-all, you’re reduced to clicking on each stroke. Ink can’t be grouped with non ink objects but must be treated separately… amazing how much this makes ink useless. And the morons wonder why the tabletPC is not doing so well.

Virtual linux

April 29th, 2008

I’ve been using linux as my main platform for ages and ages … but this doesn’t live in harmony with new laptops especially tabletPCs. Typically on getting a new laptop I had about a month of head-banging frustration to look forward to as I tried to get the power management, modem, wifi, external screen, bluetooth, and of course the tablet pen all to play nicely. Not good.

Plus, there are two applications that are worth running in windows: pdfannotator and mindmanager, both of which make a tablet useful.

So for my next tablet (a Fujitsu T4220) I decided to try a different approach. That, and I also wanted my screen mirrored on an external LCD and a projector at the same time, but upside down on the tablet only, and I wanted to turn my whiteboard interactive via something like ebeam …. I was pretty sure it wasn’t going to happen easily with linux. So inspired by Charlie Stross, I down-graded to XP grabbed virtualbox and installed kubuntu as a virtual machine.

So far it sort of works. Well it’s quite good actually. The main problem is stability, XP seems to last about a week before needing a reboot (things get pathetically slow or simply crash). On the other hand, what this setup buys me is being able to run all the peripherals without hassles and to easily switch to using pdfannotator or mindmanager without a reboot.

Most of the time I ignore XP and live happily in my virtual machine…

Thought sprints

April 21st, 2008

Here is a cool idea for startup weekends in science by Paweł Szczęsny. Why not combine the ideas of code sprints and distributed teams for doing science? i.e. set up an intense interaction for a short period for a group of researchers with the aim of say, writing a paper, in a week?

Shorter periods would be incredibly useful too. From experience, nearly every paper I have written tends to stall at some stage and lately I’ve gotten a lot out of closeting together some of the authors and just doing an extended period of discussion and writing without distractions a la a code sprint.

Imagine being able to do this online in an effective way - call them thought sprints. It would just be awesome.


Inspirational talks

April 9th, 2008

Can’t believe I missed this one: www.ted.com - TED stands for technology, entertainment, and design. It’s an annual conference where some awesome speakers are challenged to “give the talk of their lives in 18 minutes”. The talks are available for free at the website,

My favorites so far:

Dancing quantum cats

March 12th, 2008

You need to a flashplayer enabled browser to view this YouTube video
Adventures in Phase Space