If there is something you need me to take care of urgently, please let me know. Right now my priorities are with moving out of the Netherlands for a year and possibly more (which is not trivial). In some spare time I try to work on ChatZilla accessibility (specifically, my long overdue halfway report) and some of the bug reports about Chrome List that I’ve been getting. There is also Venkman demanding some attention. If anyone would like to help out with that, please have a look at the wiki page that Alex et al. created. You can usually find either of us on IRC if you’re interested in helping out. It’s really quite necessary because in its current state it is not doing that well (several basic features are broken with Firefox 3(.0.1) and on trunk).

Second, in connection to this thesis thing:

Dear Lazyweb,

I would like a way to make LaTeX do a [citation needed] in true Wikipedia style. I often find myself thinking “if I want to write something like this, I need a reference to back it up”, but searching for a good one for minutes/hours on end breaks up my writing process, so I’d much rather make a note to self and keep writing, especially for things which I am very sure of there must be a reference for. Example: “the POSIT algorithm was conceived in the 1990s by John Doe[citation needed]“.

Yes, I know that by default, LaTeX will do a [?] if I try to reference something that doesn’t exist, but that’s (a) not obvious enough for my taste, and (b) it might be the result of a misspelled BibTeX id or such, and I’d rather not mix the usecases of “deliberately left blank” and “oops, misspelled the citation identifier and/or forgot to re-run BibTeX after adding citations”. I haven’t found an easy way to do this.

Thanks!
Gijs

SightCity

I recently got back from visiting SightCity in Frankfurt am Main, Germany. The conference experience was excellent, and I had a great time with Steve, Ben and Marco, all of whom have written more pleasantly and/or extravagantly about our experiences there than I ever could. So this is all you’re getting from me, here.

ChatZilla Releases

ChatZilla 0.9.82 was released, quickly followed by ChatZilla 0.9.82.1 after a couple of nasty regressions. Almost all of the releases’ features and fixes fall squarely into the “polish” bin (eg. dragging tabs for channels named only “#” now works, and doubleclicking a user in the userlist opens a query tab, some minor accessibility and localizability fixes, etc.), except for one: CEIP, short for Customer Experience Improvement Program. Customer not being very apt, I suppose, but it’s a standard name for what is essentially a data collecting tool.

Data collecting?! Yes, data collecting. I wrote a Privacy Policy about what we do. Please read it first, before flaming me/us. Really, read it – it’s quite short! As you will have read by now, I hope, we ask first if it would be OK to collect anonymous statistics. Without consent, ChatZilla doesn’t save anything you do, nor (consequently) does anything get sent. We never collect anything personal, and both in the policy and in the UI, which is accessible from the Help menu in ChatZilla, we indicate a bunch of things we specifically won’t collect. So what can we collect, then, and how is it useful? Well, examples include the length of sessions (do people run ChatZilla for days at a time without interruption, or only five minutes?) and how tabs are handled. We can already, after just a few days, see that many people seem to be closing lots of network tabs, which may lead us to prioritize bug 249188, for instance.

So I hope that this post helps clarify that we’re not turning evil. If you disagree, and had already turned it on, you’re free to turn it off again, at your leasure. Do let us know what you think we’re doing wrong, though.

Venkman trunk issue

If you’re using Venkman with a Gecko trunk product (eg. Firefox 3 RC1, Thunderbird Shredder 3.0a1, …) you may have found that viewing source code stopped working a while ago. This happened because of a change in the way unprivileged content, like the source view itself (which is plain old HTML) is allowed to access chrome content (like the stylesheet for the source view, unfortunately). Fixing it properly is not trivial. Right now, I have suggested a more or less wallpaper fix, because I am too busy to do something nicer, and it doesn’t seem like anyone else is willing to go and fix it instead. This basically allows unprivileged content to access the chrome content again. A better solution would be to channel the stylesheet through the jsd protocol. If anyone wants to step up to the plate and fix that, that’d be awesome. In the meantime, the wallpaper patch is waiting for review. If you’re in need of a working Venkman, I uploaded an XPI to bug 428848.

Misc

I’m nearing the end of my BSc degree. I’m working on my thesis at the moment, and finishing off the two courses that remain. Perhaps I’ll write more about the thesis once I have something I can demo or screenshot in-action. For now, I’d just like to happily announce that I was conditionally accepted into the 1 year MSc Advanced Computing course at Imperial College, London. So, if you know of a good place for a grad student to live in central London (South Kensington), let me know!

At my university, I participated in an honours course about practical applications of Ethics. We covered a multitude of subjects, including (in random order) the use of marking DNA in criminal investigations (ie, putting out notices saying “based on the DNA found on this homicide victim, we think the perpetrator was black and had brown eyes), Food Ethics (what rights do animals have, should we do something about the bio-industry, and if so, what) and euthanasia: what is the right to die, do we even have it, and what are its limits.

We had guest lecturers for all these subjects. For euthanasia, our guest lecturer was Judge Eugene Sutorius, who used to be a lawyer defending doctors in assisted suicide cases back in the 20th century, and helped push the courts and the Dutch government to legislation. Currently, doctor-assisted suicide is legal in the Netherlands, if and only if:

• The patient repeatedly and explicitly expresses the desire to die.
• The patient’s decision is made well-informed, free and enduring.
• The patient is suffering from a severe physical or mental pain with no prospect of relief.
• All other options for care have been exhausted or refused by the patient.
• Euthanasia is carried out by a qualified physician, who must verify all the above.
• The physician consults one other physician, who must verify all the above.
• The physician informs the local coroner that euthanasia has occurred.

In class and small groups we discussed whether euthanasia should also be possible for people suffering from Altzheimer’s, for example – there are currently euthanasia declarations available, which some people sign, saying they want to be euthanized if they are so demented they don’t recognize their close relatives anymore. Whether or not this is reasonable is debatable, given that it’s questionable if people in such a state are even the same person they were before, and hence whether they had any right to condemn this ‘new person’ that they have become to death.

This question is certainly interesting, but not the reason I’m writing this. Sutorius explained that on one occurrence he and some others had been asked to come to the US where one state (looking at the BBC article, it was apparently Oregon) was approving legislation very similar (and in fact, partly based on) the legislation we have in the Netherlands. The writers of the bill wanted them to (honourarily or whatever) co-sign the legislation. Sutorius then explained he refused: due to the current American medical insurance system, a lot of people do not have medical insurance. He gave the following example: what if you’re a 70-year old male, you’re suffering from Multiple Sclerosis. You have 5 children and no medical insurance. Now, given the nature of MS, there is a reasonable case for severe physical as well as mental suffering, and while there is plenty of medication available, there is definitely no complete cure possible at this time. Given enough medication you can have a reasonable life for an extended period of time, but that’s about it. You “buy time”, as it were.

The problem, Sutorius describes, occurs when such a patient starts doing some maths. In a year or two, he will have used up his savings. In another five or so years, he will have used up his children’s savings. And then he will die. And suddenly, cheap euthanasia starts looking like a nice alternative. He’s had his share of life, and he’d be happier if he didn’t ‘kill’ (oh, irony) his children financially. So then, euthanasia becomes an equivalent alternative to available treatment, as opposed to a final call. This is wrong. People should not die because they can’t afford their medication.

I agree with Sutorius – and this is why I am wary of “Dr. Death”. Sure, good euthanasia legislation is necessary, and doctors shouldn’t be prosecuted if they did the right thing given their patient’s medical condition and wishes. But this should only occur given appropriate medical insurance for everyone, so euthanasia does NOT become an alternative to treatment, instead of a final call after all treatment has failed or proved useless.

The University of Amsterdam fortunately has a digital library you can access over VPN which allows you to search in lots of major databases. So I duly searched for:

“future” in Title AND  “Computer Mediated Communication” in All Words (after some failed other searches), in the “Informatiewetenschappen” subject area (“Information Science”, basically)
5th result from JSTOR:

“Promising the Future: Virginity Pledges and First Intercourse” by Peter S. Bearman.

No comment.

In other news, I went to see Anna Ternheim in Lantaren-Venster yesterday. This was fun, though the dual concert approach wasn’t all that successful in my eyes – Vinicio Capossela simply makes a very different kind of music, so to me this didn’t better the experience much. I’m not sure if, in hindsight, I should have preferred missing his bit of the concert. Regardless, I liked the somewhat more intimate approach (about 200-300 people attended) than what you generally see on tv (this was the first ‘pop’ concert I ever attended – I’ve been to classical concerts before). I had a good time, though I unfortunately didn’t manage to get an autograph on the cd I bought, which is sort of a shame. Oh well.

Edit: You can listen to some of her music on MySpace. I wouldn’t normally link there, but this is the only place I can find where you can listen to “No Subtle Men”, which she sang last night and is absolutely great, if you ask me.

• Trying to find time to co-write a paper and to get it accepted for a student conference on AI in Utrecht.
• Trying to find time to modify a paper on Ethics (specifically, Moral Relativism) that I wrote some time ago and submit that to ’something’ Dutch / UvA.
• Doing regular coursework, which involves
  • reading about 200-300 pages every week
  • writing summaries and questions for speakers on the subjects for all of them
  • working on a project with students from Stanford University and a Fortune 500 company in the tech industry. Can’t say much more than that, though for once it doesn’t involve engineering.
  • writing language processing tools in Python.
  • trying to catch up with lost classes from last block about Statistics and Stochastics.
  • writing knowledge processing tools and representations using Prolog and OWL.
• Working on my Mozilla project. I meant to write about this earlier, but the Mozilla Foundation has sponsored me to work on making the ChatZilla IRC client accessible (as is clear from that link, it also needs a better website, we’re working on that on the sidelines). In fact, I have a conference call concerning that starting in 10 minutes, so I’d better finish this up.
• Working on other Mozilla-related issues
• Trying to get some work done on some other Firefox extensions of mine.
• Figuring out what university I’d like to attend for my MSc.
• Visiting junks drug and alcohol addicts. We need to stop calling people garbage – things are garbage (see also the post title) but people never should be. Also, Amsterdam is strange in the sense that I just went to a church service done by the Drug section of the pastoral care (not sure if that’s the right way of putting it in English) and atop of us (ie, in the room above us) some snobby people were having champagne, for the opening of some exhibition or whatever. Stark contrast…

That will do for now. Off to that conference call!

I’m not sure who made this joke up, but it’s not funny. I already blew at least one of the courses I had the first part of this semester (because thanks to United Airlines I missed the exam), and now the schedulers have been conspiring to make me fail more of them this part of the semester. Oh, and I don’t know if this is just me, but there’s this day called “Friday”, which you can actually USE if you’re having trouble scheduling…

• Ik heb m’n honoursmodule over Ethiek goed afgesloten, en ben nog bezig te proberen mijn essay daarvoor gepubliceerd te krijgen, daarvoor al een gesprek gehad met een “echte” ethicus enzovoort. Leuk allemaal.
• Nadya, Hylke en ik hebben onze honoursmodule binnen de Kunstmatige Intelligentie Bachelor bijna afgerond. We moeten ons rapport nog wat redigeren en een presentatie geven, dan zijn we klaar.
• Ik ben begonnen aan een minor Sociale Informatiekunde, wat gaat over de manier waarop mensen computers en informatie gebruiken, en hoe dat verbeterd zou kunnen worden. Dit is best interessant, alhoewel het eerste vak dat ik heb me tegenvalt: weinig nieuwe theorie en de praktijkonderdelen zijn mij niet uitdagend genoeg, te ‘gespreid’ en toch te veel werk. Er wordt verwacht dat je er 15 uur per week in steekt. Je moet de opdrachten in groepjes doen, maar je hebt naast het hoorcollege maar 15 minuten per week samen ingepland staan in het rooster (en bovendien zijn voor mij zowel het hoorcollege als het samenzijn dubbel ingeroosterd met Kunstmatige Intelligentie vakken). Het is dus veel werk, maar wordt gespreid over een week (in plaats van op projectbasis een paar dagen achter elkaar bikkelen). Bovendien is het niet uitdagend genoeg in die zin dat de verwachte resultaten (paper prototypes, bijvoorbeeld) bijna geen werk vragen en ik niet zo gemotiveerd ben om er uitgebreid mee bezig te zijn als het resultaat zo onsubstantieel is.
• Ik probeer erachter te komen wat ik van mijn opleiding verwacht en wat ik als Masteropleiding zou willen doen (en waar dan). Wat mij tegenvalt aan KI is dat van de dingen waarvan mensen zeggen dat we ze kunnen (spraakherkenning, gezichtsherkenning, emoties, natuurlijke bewegingen, kennissystemen, etc. etc.) we de meeste dingen in werkelijkheid niet kunnen, gewoon omdat we er geen goede manier voor hebben, en dat we het vervolgens toch doen met gebruik van allerlei flauwe ‘hacks’ die in zeer specifieke gevallen het probleem oplossen. Het makkelijkste voorbeeld is kennissystemen in combinatie met natuurlijke taal: er zijn op dit moment systemen op de markt die bijvoorbeeld dokters kunnen helpen met het maken van diagnoses, en systemen waarbij mensen door middel van spraak een computer bedienen. Echter, dit zijn heel specifieke taken. De computer begrijpt je niet als je over een ander onderwerp dan medische diagnoses gaat praten (het weer, wat je voor het laatst gegeten hebt, etc.) of als je woorden of termen gebruikt waar de computer niet bekend mee is. Dit kunnen deze systemen over het algemeen niet leren, en op die manier vindt er nog steeds geen ’sterke’ KI plaats, wat je wel graag zou willen…
• Waar ik een masteropleiding zou willen doen: de meeste goede universiteiten die iets met KI doen zitten in de VS. Dat is best ver weg, en de VS als land trekt mij absoluut niet. Uit mijn vakantie-ervaring, en de inwoners ervan die ik heb leren kennen, zou ik liever naar Groot Brittanië of Zweden gaan, maar geen van beiden heeft KI opleidingen die als zodanig zijn aangeschreven (in plaats van vakken / ‘tracks’ bij “Computer Science”, waar ik waarschijnlijk weinig tot geen nieuwe dingen zou horen). Moeilijk, moeilijk…

Hierboven zie je een cilinder op een gevlakt oppervlak. Dat ziet iedereen meteen vanwege de schaduwen, het perspectief van de vlakjes, enzovoorts. Stel, we willen de computer in ieder geval onderscheid laten maken tussen de lichte en donkere vlakjes. Dat is niet zo moeilijk, want al die pixels hebben gewoon een hoge of lage numerieke waarde voor de computer, afhankelijk van of ze licht of donker zijn. Dat valt best tegen: klik maar eens op het plaatje. Die streep die je ziet heeft de kleur (‘donkerte’) van het lichte vlak in de schaduw van de cilinder, en ook de kleur van het donkere vlak bovenaan! Klik nog maar eens als je niet gelooft dat het hetzelfde plaatje is.

Waarom zien onze ogen dat verschil wel zo duidelijk? Nu, we zijn gewend aan het compenseren van schaduwen, diepte en andere lichtverschillen. Dit is ook de oorzaak van allerlei andere ‘foutjes’ die mensen maken als ze naar bepaalde ‘getructe’ plaatjes kijken.