Tuesday 18 December 2012

Dear UPenn Engineering, Please read my open letter to Judge Business School

Dear U Penn Engineering

Like Judge Business School, you use a third party site to obtain your references for applicants to your graduate programs.

The appalling nature of this software, its lack of usability, displays that it seems you hold in contempt the time and energy of the people providing a service to you by writing references.

I've ranted about this before, so I refer you to that open letter.  Cambridge quickly got back to me to say they would reconsider their processes, so I hope you will too.

One differences from Cambridge.  First, I don't know the data protection provisions in the USA so you may ignore that part.

But here's a new point which was true of Judge too, but only just occurred to me.  Why did I get a request from a great University like U Penn, with zero contact information for U Penn?   The only contacts are for the firm supplying the reference software.   But maybe I would like to find out what the programme is, and a point of contact if there is anything I need to discuss with a staff member at U Penn for any reason.    Shocking.

A final oddity.   Having created and uploaded headed notepaper, as requested, when I preview the document it is overwritten, making both my University crest look ugly and the overprinted information illegible.

Best Wishes

Ian Gent

Friday 14 December 2012

Asking for a pardon for Alan Turing alone is a shameful act

Today, a number of famous scientists have published a letter in the Daily Telegraph  requesting a posthumous pardon for Alan Turing's conviction of homosexuality. I think this is a shameful act, and these great scientists should have known better.

A picture of Alan Turing running.
Turing once outran an Olympic Silver Medal runner
I have very mild tangential connections with some of them, very very mild.  I met Doug Gurr a few times while he was doing his PhD at the same time as me, before he moved on to greater and better things.  Tim Gowers is a friend of a friend.  And my father Hubert Gent and Martin Rees were in the rather small UK Radio Astronomy community in the 60s and 70s.    Oh, and Lord Grade worked with Morecambe and Wise, who both I and my daughter love.   But I hope that if I had very close connections with all these people, it would not stop me criticising them for asking for a pardon for Turing.

You might think I'm against retrospective pardons and apologies, but that is not my point at all.

And before I go on, let me say I am a huge fan of Alan Turing.

I work in Artificial Intelligence and Computer Science. When lecturing on Artificial Intelligence, I often give my reasons for regarding Turing as my hero.   He was a key founder of theoretical  computer science with the Universal Turing Machine.  He was a key founder of practical computer science, with his work on the Bombe during the war and on the ACE machine afterwards.   He was a founder of theoretical Artificial Intelligence when he wrote about what became known as the Turing Test. And he was a founder of practical Artificial Intelligence when he wrote the first chess program for the full game of chess, which in fact had to be run using pencil and paper as it was too sophisticated for existing computers.  Oh, and in between all these things, he played a major part in winning the war with his codebreaking.

I regard him as one of the three greatest British scientists of all time.  I think Newton and Darwin and him are the great three, and would not wish to choose between them.   There have been many other great scientists, but these three essentially founded major world disciplines, as well as having the most astonishing scientific insights.   When I became professor at St Andrews, the University gave me a book as memento, and I asked for Turing's collected papers on Artificial Intelligence.

If you still don't believe I'm a Turing-fan, let me tell you my one and only twitter username is turingfan.

After all that, why do I think nobody should ask for Alan Turing to be pardoned?   Actually, I don't think that.  I believe - very strongly - that it is shameful to ask for a pardon for Alan Turing alone.   Alan Turing was convicted of what was then a crime, and I don't believe there is doubt that he was guilty.   Many many other people were convicted of similar crimes which would not be crimes now.  All of them deserve pardons or none of them do.

Why should Turing benefit from a pardon and none of the others? Because he was famous?  Because he was a great scientist?  Because he helped win the war?  Are any of these reasons for pardoning somebody for a real crime?  Of course not!

The reason to pardon Turing is  that we now believe homosexuality is not criminal and that it should not have been in the past.   If we believe that - and if we believe that retrospective pardons are useful - we should pardon everybody convicted of crimes of homosexuality which would not now be criminal.   To single out Turing to benefit is shaming, and I'm sorry to say that with most of the signatories of the letter being scientists, it is shaming to the scientific community.  

If you must ask for a pardon for Turing, there is only one way to do it. Ask for a pardon for ALL those convicted of similar offences, Alan Turing and many many others.   Use Turing as a poster boy example if you wish, but never ask for a pardon for him alone.

There must be many people alive now who were convicted of offences related to homosexuality.  They all deserve a pardon.  They would even benefit from one.  Ignoring them and asking for a pardon for Alan Turing is shameful.

Tuesday 4 December 2012

Dear Judge Business School, Cambridge: Whatever your online reference guys are paying you to use their software, it's not enough

Today I filled in a reference for a former student who wishes to undertake a course at Judge Business School, University of Cambridge.   The experience was so bad, so bad, that it inspired me to write this blog post.  

Let's remember the setup.   I'm not being paid to write this reference.   I'm doing it as a favour to my former student, and as a favour to my academic colleagues at the University of Cambridge.  I'm happy to do it to help both sides.   But - and it's appropriate to repeat this when it is after all a business school I am talking about - I'm not being paid for this task.

Also let's straight away take the student off the hook.  They want to do a course at this famous Business School, so they have no choice but to ask me to supply a reference according to the demands of Cambridge University.

You might think in these circumstances - doing unpaid work for their direct benefit - the University of Cambridge might make my life as easy as is reasonably possible.

Instead they use some online reference software provided by - I won't name them because they don't deserve an extra hit on Google Search.    Though now I think about it, the name of the firm is appropriate as it reminds me of Hobson's Choice, an unpleasant choice where no alternative is offered.  

Here are some of the highlights of my experience. Ignoring of the course it being ridiculous to have to use special purpose software in the first place - Cambridge should give me multiple options to make my life easier.   But let's pass that by quickly.

First piece of joy.  I was sent in plain text a username and password.   (Yep, password in plain text because nobody at Cambridge is a world expert on security).  

Well, actually that's pretty standard, it washes by me.  But first thing I had to do was ... be forced to change my password!  For what is to all intents and purposes a one-time logon: it would be a miracle if this login was used the next time they ask me to write a reference.   As usual, non security enhancing requirements were made.   So yeah, they wasted my time making me choose a password I will never use after they sent me an initial password by email.

Another thing I particularly liked was being asked to provide a phone number with numbers only.   Guys, phone numbers sometimes require non numbers, especially the + for country code.  And people like to write spaces and brackets while computers are good at stripping them out.  But the better one was being told the date format for today was incorrectly entered: I think I left out the initial 0 on the date.   It would of course be hard to automatically enter today's date for me.

Apart from starting to mentally write this blog post at this point, I actually got to the bit where I could start to enter my opinions.   Which was not too bad except that ... after being forced to use this web form I was not allowed a text box to enter my comments in.  I  had to upload my comments in a document.   Umm, why no choice to type it or paste it?   I was asked to do it on headed notepaper, because umm, actually I can't think of a reason.   Headed notepaper used to be a very slight fraud deterrent, indicating the letter was from somebody with access to it.  Nowadays of course, it's a mild pain for academics at a University to provide an electronic document on headed paper (the clue is in the name, paper, this is an electronic document guys!)  and of course no problem at all for somebody who intercepted the email to forge if they wish.

All of this shameful non-usability, inconvenience, and complete lack of consideration to the person who is paying with their time and effort, is sadly par for the course in writing references these days.  I probably wouldn't have bothered with this post except for one last vignette.

At one point I was told that my comments were confidential and would not be disclosed to the candidate.    So apparently Judge Business School (or their online reference providers) have some magic exemption to the data protection act?   The irony is, that according to my University, I cannot be forced to provide this reference, but Cambridge might be.   Actually I always assume that references might be shown to the student, and would never write something that fundamentally could not be (even if what I write may not always be what they want to read.)   But a blanket statement which essentially ignores the Data Protection Act - and remember provided by a company who is providing a service so are assumed to be experts?

I titled this post by stating that whatever these choice people are paying the University of Cambridge for the right to upset their referees and thereby disadvantage both the University and their potential students, it's not enough.   I take that back. If it's several million a year then that's enough to pay me £1000 or so for the benefit of my reference and my later thoughts of this blog post.   You've got my address in your reference system.

best wishes

Ian Gent

Monday 15 October 2012

Understanding the Nobel Prize

It's not often - maybe never before now - that I've understood technical aspects of the work behind a Nobel Prize.

But today I found out that Shapley has shared the Nobel Prize in Economics with Al Roth.  Sadly Gale has died so cannot be awarded the prize.

The crux of the work is the Gale-Shapley algorithm on "stable marriages."  So yes, deep down the Nobel prize has been awarded for an algorithm.

The idea of a stable marriage is when men want women and women want men, but we don't want marriages which will break up.  A marriage will break up if there is an "unstable pair", two people who would rather be with each other than their spouses.   It's all theoretical - it doesn't assume same-sex marriages, polygamous ones, or romantic preferences that change with time.  But it leads to a lovely algorithm, and interesting insights.  It's really interesting stuff and lovely.  I've written some papers on it from the point of view of constraint programming, my research area.

11 years ago I co-wrote a paper on this topic for the big conference in Constraints.   It was a privilege to work with Rob Irving and David Manlove.   And I don't think that Barbara or David or me will mind admitting that in that case the key authors of the paper were Rob and Patrick Prosser.   The thing that interested me on the topic was that it turned out we could turn the efficient Gale-Shapley algorithm into an efficient constraints algorithm (in a certain technical sense, no need to go into that here.)    But what was even better was that this technical efficiency turned out to be interesting in its own right, and analysis of it led to a big part of a PhD, specifically Martin Green's to be precise, Holloway, 2005.   Not the only PhD in constraints and stable matching neither - here's to Chris Unsworth -  but maybe the one that took our work in an unexpected direction.

However for practical purposes, our work in Constraints didn't really take off.  We hoped it might, but it didn't.  The real practical purpose of this work is kidney-matching.  This is a huge ethical problem for which people quite rightly (in most people's view) don't think money is the right solution.  The solution is a stable matching where everyone benefits.   That's what part of this Nobel prize for, for Al Roth.  And that is what David Manlove uses this work for - of course extending it as well for scientific interest, but also helping people get kidney transplants they wouldn't otherwise get.

So this is what a Nobel Prize means if you understand something about its technicalities.  It turns out it has influenced people many decades later in other countries, and then their work influences other people in turn, in a research area some way off the original.  And then people get new kidneys.

p.s. Yes I know it's not really the Nobel Prize, it's the Nobel Memorial Prize, but everybody calls it the Nobel Prize in Economics including the BBC and the Guardian.

Tuesday 7 August 2012

Not the way the world's greatest sportest event should be

Thanks to Neil Moore for this link. The article quotes one Prof Forrest: "We have identified four sports where there is virtually no chance that anyone from a poor country can win a medal - equestrian, sailing, cycling and swimming"

Meanwhile today the UK just won a gold medal in triathlon which combines two of those sports (plus running.) And celebratory BBC news article says "Further medals could follow on Tuesday in cycling, dressage and windsurfing." Which are - you guessed it - three of the four rich-country sports. With the triathlon gold partly for swimming to complete the set. 

I am genuinely happy about British success but at the same time very saddened that poor countries are eliminated from competition in many key areas of the games. Not the way the world's greatest sporting event should be. 

Athletics is more or less an exception. If you can run fast you can make enough money to compete at the highest level wherever you are. And critically be identified early enough to be trained well enough in the USA. So I think it's wonderful that the 10000m champion was born in Somalia, grew up in the UK and ran for Britain, and lives in Portland Oregon.

Tuesday 14 February 2012

Shooting a googlewhackblatt in the foot

A googlewhackblatt is a single word query to Google that gets one hit.   And one such word is autopedoballistic.   Made up to mean shooting yourself in the foot.   But since I have just posted this page, I am guilty of autopedoballistics by ruining it myself.  So here's the proof.