28 September, 2004: You scratch my back, and I'll bomb yours
So, there was an interesting piece in the Economist this week on the funding of political parties, which mentioned the splendid Fund Race site which lets residents of the United States determine how much cash their neighbours have given to America's two political parties. It produces beautiful maps liks this one:
showing donations by individuals and companies overlaid on street maps. No doubt this makes for interesting discussions in the elevator.
When I first saw the Fund Race site, it occured to me that it would be interesting to do the same thing for the UK. Here, data about political donations is held and published by the Electoral Commission under the Political Parties, Elections and Referendums Act 2000. (Parties themselves are registered under a 1998 Act; I have to say the idea of the state acknowledging the existence of political parties makes me uneasy, but since I'm also in favour of state funding for parties it's clear that my views in this area could do with a little tidying up....)
However, for reasons of privacy, donations by individuals to the various parties -- although recorded -- are not published with their addresses, so it's not possible to map individual, personal donations as with Fund Race. Donations by corporate bodies are recorded with their addresses, but the results aren't very informative: (these show the data for 2003; there are fewer circles than donations, because I've amalgamated circles which overlap too much to be clear)
And before anyone complains, yes, I know the circles should have areas proportional to the sizes of the donations (rather than proportional to the log of the donations, as here). But doing it `properly' gives ridiculous results -- firstly, because the Labour party received so much money than all the other parties look insignificant in comparison -- Labour got £14 million, as opposed to about £4.4 million for the Tories and £1.4 million for the Liberal Democrats, not counting public money and donations from trusts; and secondly, because so much money is donated from addresses registered in London that all you see is a great big circle obliterating the south-east, plus scattered pustules across the rest of the country -- probably a bit like the maps Soviet war planners would have drawn, I suppose.
Actually this whole exercise is a bit pointless because the UK is so centralised. Many Labour party donations, for instance, come from trades unions whose headquarters are in London, even though their members work all over the country. Similarly, Tories get donations from big companies (boo!) which are mostly based in London. (Actually Labour seem to get more donations from big companies....) Liberal Democrats don't really seem to get many donations at all, but many of them come from Liberal clubs all over the country, which I suppose is a bit more reasonable.
The Economist article also suggested that,
Perhaps giving, like voting, should be anonymous? Voting was actually public in America until the end of the 19th century. Ian Ayres, a professor at Yale Law School and co-author of Voting with Dollars, thinks the answer is private donation booths; these would prevent candidates from knowing whom they owed special favours, just as secret ballots do. But if campaign gifts were anonymous, there would probably be fewer of them.
The point here being that you might give a large sum of money to, say, a British political party, in return for which that party, when in government, might (for instance) not ban tobacco advertising in your sport, or buy a large quantity of a probably-useless pharmaceutical product from your company; or you might give a large sum of money the an American political party, in return for which a future president from that party might (for instance) invade the Middle Eastern country of your choice or let you drill for oil in an unspoiled wilderness. And even if the bribery did not work -- and, for the avoidance of doubt, all of the above examples are, of course, completely hypothetical -- the perception of bribery is itself corrosive, and therefore to be avoided.
The secret ballot is intended to stop intimidation of voters by candidates, by making it impossible for voters to prove how they have voted. Similarly the idea here is that, if it is impossible for a party to find out who gave them their cash, they won't know what they've been bribed to do. (Note that I'm -- naturally -- assuming the basest of motives here.)
Unfortunately this won't work. In the donation case, both the donor and the party have an interest in finding out who's responsible for which donations (the donors, so that the party is given the correct incentives, and the party, so that it knows what it's being paid to do). And they have rather a lot of information to go on, which must be recorded because the party has to keep its accounts in order. In the simplest case, I can give a party an arbitrary sum of money of about the right size (say, £3,080.17 or whatever), and then ring them up and say ``I am the donor who gave you £3,080.17 on such-and-such a date, and I'd like you to do the following favour for me''. You might think that you can close down the hole by not reporting donations to parties individually, but funnelling them through some external agency which adds the amounts together, but that just makes the task harder, not impossible. For instance, if every week's donations are aggregated, I can give £x in the first week, £y in the second, £z in the third, and so forth, and report the values of x, y, z, etc. to the recipient, who can compare them to the totals in each week and -- if the values are chosen carefully -- determine with confidence whether the donation was actually made or not.
Some readers will spot the analogy with, for instance, this story about `covert channels' in the design of MULTICS; it's interesting if a bit technical, but the point is that if two parties can exchange any information (however limited the amount), it's possible for them to communicate with one another. (Note for the pedants: this hypothetical example isn't quite the same as the MULTICS one.) I believe that the same sorts of problem have arisen in the design of (e.g.) spectrum auctions to telecoms companies.
I don't think that this is an argument for state funding of political parties, but it's certainly an argument against one of the alternatives.
27 September, 2004: Shite on sight
Disappointingly few people answered the one-question quiz in my previous piece, and all but one got the answer correct. (Apologies, by the way, for the long delay in writing this piece, which has rendered it mostly out of date.) Nick summarised the reasoning -- such as it was -- basically correctly; Pete's answer was imaginative, but for once national-security idiocy trumped intellectual-property idiocy. The (alleged) TERRORIST, was, of course, the chap who dressed up as an Abu Ghraib torture victim and protested outside a US Army recruiting centre; the Massachusetts Police arrested him for,
disturbing the peace and felony charges of making a false bomb threat and using a hoax device. The charges apparently reflect the District Attorney's concern that Mr. Previtera might have been mistaken for a terrorist.
`Apparently' indeed. If anything the charges reflect the fact that the Attorney General of Massachusetts is an ass. But I think we already knew that that is no disqualification for becoming the Attorney General of an American state. Thank goodness that would never happen here.
When writing the quiz I should have added the anti-hunting protesters who recently found their way into the chamber of the House of Commons to the multiple-choice options, but unfortunately I wrote the piece before they'd done the deed. Meanwhile I am entertained to see that while numerous MPs have been complaining that this is the greatest violation of the privileges of the Commons since Charles I attempted to arrest five MPs in 1642, none of them should have mentioned the Luftwaffe's success in bombing the place out in 1941. I suppose that in the interests of European unity it's best that they didn't.
So, lately, there have been several shocking security breaches in this country. A protester dressed as Batman(R) got onto a balcony at Buckingham Palace; his accomplice, dressed as Robin(TM), didn't, because a police office threatened to shoot him unless he got down off a ladder. So he did. I don't know about you, but I don't much fancy living in a country where police officers routinely threaten to shoot people who don't obey their commands, but I suppose now that we're `at war', we have to put up with this crap.
Secondly, anti-hunting protestors got into the chamber of the House of Commons, provoking more calls for a shoot-on-sight policy (as opposed to a get-threatened-by-man-wearing-tights-and-a-sword policy, which is what the House of Commons presently has). And a journalist from the Sunday Times got into Holyrood until he was challenged by a builder. I don't think that anyone's called for a shoot on sight policy to apply to journalists from Sunday newspapers, but I don't claim to follow these things rigourously.
Calls by journalists (or, worse, other web-loggers) for shoot-on-sight policies are, I suppose, best interpreted as evidence for a slow news day rather than (for instance) any actual qualifications for prescribing new security policies, but for those who may be tempted by these ideas, ask yourself the following question:
Suppose that guards outside the House of Commons are armed and given instructions to shoot any person who does not promptly show the proper credentials. Who is more likely to be shot? Is it (a) a suicide bomber; (b) an anti-hunting protestor; (c) a cleaner, absent-minded MP, visitor or other innocent person?
Give up? Well, a clue: it's not (a) and it's not (b).
Coincidentally this is the same problem as expected for proposed biometric security screening at airports (in particular, for the laughable `facial geometry' thing proposed for new passports).
(And one other thing: I don't normally do these `Google bomb' things, but I suppose I may as well direct you to Harry Hutton's splendid killer fact about the British National Party, which might cheer you up.)
14 September, 2004: If you answer incorrectly...
... then THE TERRORISTS will win. Since my last quiz was quite successful, here's another (shorter and easier) one:
Which of the below people is one of THE TERRORISTS?
| Dave Pyke | Joe Privetera | Luis Posada Carriles |
|---|---|---|
 |
 |
 |
| Dressed as `Robin' from `Batman', Dave Pyke gives an interview outside Buckingham Palace shortly before his arrest for involvement in a protest by pressure group Fathers 4 Justice. | Joe Privetera, protesting against torture by US forces in Iraq, outside an army recruiting center in Boston in the United States. | An old photo of Carriles, who was imprisoned in Venezuela for murdering 73 people in a 1976 airliner bombing; he's also implicated in six hotel bombings in 1997, and in an attempt to murder a famous head of state. He was recently pardoned and released from prison. |
As with the estimation quiz, it's pretty easy to cheat on this one. So don't.
11 September, 2004: More introspective nonsense
So, then, Slashdot. That gets you something like 50,000 hits in ten hours, and contrary to naive expectations, some of the visitors wrote interesting comments. This also confirmed something I've long believed, which is that the `Slashdot effect' is something which mainly affects people who've misconfigured their webservers:
-- from the theory, the hit rate should rise instantly from its background value to its maximum value at the time the link was made. In fact it rose in two jumps, because we'd forgotten to remove the compile-time limit on the number of processes with which apache can serve clients. Oops. Once that was fixed, the problem went away.... (Five hits per second isn't very much, of course -- sustained, it's about four hundred thousand visits/day -- but that plot doesn't include the requests for all the graphs in the two discussion pieces to which Slashdot linked; including those pushes the rate up to about 35 hits/second.)
Slashdot didn't link to the quiz itself from the front page, which is slightly disappointing given that the linked discussion articles give away many of the answers. That said, judging by the Slashdot comments, I don't think this will have had much effect on the results....
One other oddity: numerous people complained that, in relation to the question about the first year in which a woman flew in space, the term astronaut is specific to American spacefarers and that the first woman in space, Valentina Tereshkova, was a cosmonaut. This is a perfectly legitimate point; such framing problems will affect the answers given.
The question doesn't even mention the term `astronaut'. It reads,
Please estimate... the year in which a woman first flew in space.
On a slightly different topic, here are the quotations and aphorisms which the Slashdot hordes liked most and least from Am I Sig or Not?, based on the change in score from before they arrived to after:
The best that can be said for this, I think, is that the downward-moving ones don't confirm our prejudices about Slashdot readers as much as do the upward-moving ones.
9 September, 2004: Niel Humphreys of `Snowdon Computers' is a lying incompetent cunt
(Yes, it's time for another episode of `adventures in customer service'.) Occasionally I buy things via eBay, an online auction service. Usually these transactions proceed smoothly, and items usually sell well below their retail values, so it's good value. Occasionally, however, they don't.
(I'm aware that this isn't a very interesting piece, but obviously I have a duty to warn others against dishonest merchants.)
Last week I bought from `Snowdon Computers' (which appears to be a one-man operation) a terminal server (if you don't know what one of those is, you probably don't want to). This was described by its vendor as `removed from a working environment' (eBay for `was working when obtained by vendor') though the vendor didn't claim to have tested it himself. I bought this item.
On arrival, it didn't work. I inspected it and discovered that it had suffered from a large current discharge, almost certainly from a nearby lightning strike (this is a common cause of failure for terminal servers, because they are often connected to long cables which are vulnerable to nearby strikes); this current had destroyed one of the rear-panel sockets:
delaminated a PCB trace (top arrow), and arced to the device's box (bottom arrow):
(apologies for lousy photo). It is not surprising that a device damaged in this way didn't work very well, and as you'd expect about half the ICs on the board were in a short-circuit condition just sitting there getting hotter and hotter rather than actually doing anything.
So I pointed this out to the vendor, expressing my surprise that he had not seen fit to mention this obvious damage to the device in his advertisement. He said that he hadn't opened it up at all(!). I pointed out that it is fairly dangerous to be selling electrical equipment without inspecting it, and he replied (hilariously styling himself Niel Humphreys of the `eBay department', presumably in an attempt to make his pissant operation look bigger than it really is),
Yes, I do not consider computer voltages to be dangerous.
-- suggesting that the talentless fucknut doesn't really understand enough about electricity to be dealing with it in a professional capacity.
Naturally I asked for a refund, since given the above I felt it unlikely that Mr. Humphreys was competent to repair the unit. (It certainly is repairable, but not, I suspect, economically.) Initially, shit-for-brains refused, even, idiotically, remarking,
I note you are relatively inexperienced on Ebay. I can only suggest in future you read and understand auction descriptions fully and do not bid on any items if you are not willing to accept the level of testing advertised. This item was listed in good faith with no reserve according to the auction description, I did not ask you to bid the amount that you did & expected you to evaluate the risk in relation to the information given. It seems you expect the best of both worlds, perhaps Ebay is not for you?
This suggests that he doesn't understand the Sale of Goods Act either. The Act -- which covers sales via eBay just as well as it covers sales in a shop, a fact which I fear some eBay customers don't know and which some eBay vendors no doubt don't want them to -- requires items sold to:
``conform to contract''. This means they must be as described, fit for purpose and of satisfactory quality (i.e. not inherently faulty at the time of sale).
An obviously faulty item which has been destroyed by a lightning strike cannot have been from a `working environment' and so is not `as described'; nor is it `fit for purpose' or `not inherently faulty at the time of sale'. The Sale of Goods Act gives the customer (roughly) the options of repair, replacement or refund of cost of an item which is not of merchantable quality.
Although Humphreys has now refunded the money I paid him, he evidently felt that he was doing me an enormous favour in doing so. Since the worthless fucktard doesn't seem to understand the legal requirement to sell goods of `merchantable quality', I would suggest that it is him, not me, who is unsuited to using eBay.
eBay has a system of `feedback' in which you can leave your comments on the results of each transaction. Reviewing my purchase of his dodgy crap, he wrote,
Item supplied as described, returned totally blown & useless. Bidder admitted he'd had the case open and poked around inside.
This, of course, is a lie; and it is defamatory, since Humphreys is implying that I damaged the item. (I did not, by the way, `admit' that I had opened the unit: I showed the little shit the photos above, to demonstrate to him that he had sold me faulty goods which were dishonestly described.) His account is also not consistent with his behaviour; if I really had destroyed the terminal server, he would presumably not have refunded my money, as he would certainly not have been liable to do so. In fact, he knew damn well that he had to give me a refund, since basically he'd tried to defraud me and I'd called him on it. He is still trying to weasel out of refunding the cost of returning his faulty goods to him. We'll see about that....
(I was also amused that Niel Humphreys is obsessed with the phrase `poked around', sparing no opportunity to describe the simple process of inspecting electronic equipment as `pok[ing] around inside' it. Perhaps psychoanalysis would furnish an explanation of this fixation.)
So, in sum, do not buy anything from Niel Humphreys of `Snowdon Computers'; on eBay he presently calls himself `snowdonia2002'. He is a liar and incompetent; he sells shoddy goods which are dishonestly described; and he is reluctant to fulfill his legal obligations. It is often said of such people that they are not fit to run a `whelk stall'; I suspect that would be far beyond the abilities of Mr. Humphreys.
(Other than this, by the way, my experience of eBay has been good. eBay itself is an odd organisation; they are the largest online auction outfit in the world, yet they are barely capable of sending a correctly-formatted email, and their website is, frankly, shoddy. But most eBay vendors I've dealt with have been scrupulously honest and efficient. All the more reason to draw attention to those, like hopeless numpty Niel Humphreys, who are not.)
So, sorry: no graph today, though this digression may entertain you.
Update: hopeless dishonest person Niel Humphreys is apparently so self-obsessed that he couldn't resist posting about this page on a public newsgroup, writing inter alia: (emphasis mine)
URL reservered [sic.] (email me if you want a laugh & I will send it [later he posted it to the group]) as it is highly profane and I was basically in the wrong anyway but the guy was such a dick about it I decided to argue. I did refund after a bit of an entertaining barney though. :)
-- showing, I think, that the English words `basically' and `anyway' now have almost no meaning.
A commenter -- who did not follow the comments policy and whose comment is therefore no longer visible -- remarked on the fact that `snowdon2002' has (other than mine) no `negative feedback' on eBay. The commenter argued that this suggested that my complaints above about human turd Niel Humphreys are unreasonable. This is one possible interpretation, though it is not the correct one. Note from the history of pseudonyms on the eBay site, Niel Humphreys regularly changes his eBay pseudonym, presumably in order to escape from any negative feedback. (eBay do not show feedback for his old pseudonyms, so this cannot be verified.) Secondly, Humphreys is a keen student of how to remove `negative feedback' from his profile. It is hardly surprising that there isn't any there.... (There's a point in here about reputation systems which is of wider interest; perhaps I'll write about those later.)
Update #2: the silly cunt writes in another newsgroup post,
If he's have been reasonable and polite from the start he'd have found me reasonable and polite and I would have refunded without question immediately.
Here is the text of my first email to him:
Having now found time to test this unit, I discover that it has suffered fatal damage, most probably from a lightning strike. The damage is obvious from a cursory inspection; I am surprised that it was not mentioned in your description of the item. Thoughts?
(As another brief note, posting this seems to have brought all sorts of creatures out of the woodwork, all of whom seem intent on commenting but somehow have failed to read the comments policy, which states various guidelines for commenters, including that they should: write something interesting; give their full real name and email address; and avoid errors of grammar and orthography. It's no good expressing surprise that your comments don't get displayed if you don't comply! Idiots.)
(I have removed the word `crook' from the above description. I am absolutely and wholly satisfied that Mr. Humphreys is in no way a crook and apologise unreservedly for any offence caused by my so describing him.)
8 September, 2004: Police Five Dozen
A little while ago I moved house. On my way home one evening soon after doing so, I passed van after van full of police officers clutching riot gear, and when I actually reached my door, there were two dozen constables standing outside it. (I should say that this rambling anecdote is just here to assuage my guilt in just introducing a link to somwhere else. You can skip to the end if you just want to get out of here....) Of course, having just moved I didn't know exactly what my new neighbourhood was like. But I'd only moved a few hundred yards, so I reckoned I had a pretty good idea. Thus I was perturbed. I asked the nearest copper whether there'd been any trouble; clearly, the days in which a constable would automatically address anyone vaguely middle-class-sounding as `sir' are long over, since the chap replied simply, `no, we're just shutting the pub'. And, indeed, there were revellers -- well, at least, glum-looking and inadequately-dressed people -- pouring out of the boozer opposite in order to mill around on the street corner. Obviously I'd make a rotten policeman, as I'd imagined that they would be more disruptive out there than inside the nice warm pub with a pint or seven of European Fighting Lager inside them. There wasn't any significant trouble, as it turned out. Well, there wouldn't be, would there, not with a whole platoon of police looking on. At least, not here. I don't think Cambridge's residents are natural rioters; apart from a pitched battle with off-duty soldiers in the mid-'70s, I understand that the place has been fairly quiet ever since the Civil War.
I don't think I've ever seen so many police in one place at once -- not in Cambridge anyway, other than at animal rights demonstrations. The number seemed more appropriate for ensuring that demonstrators remained inside a `free speech zone' or outside an arms fair than for closing a pub early, even on a Saturday evening. The event which occasioned this police presence was not, as I had initially assumed, a tip-off that Osama bin Laden was out on piss or perhaps that the publican had hidden the Weapons of Mass Destructions behind the bar, but in fact the annual Midsummer Fair. Naturally the weather was lousy, so rather than entertaining themselves by trying to win coconuts or strapping themselves into one or other sort of centrifuge a hundred and fifty people had sensibly spent the afternoon in the pub, whence they were being ejected by PC Plod and his more menacing colleagues, some of whom were -- I kid you not -- festooned with Batman-style utility belts while others stood around filming the ejectees. I'm not sure whether this is a tactic to intimidate the mob, or if it's just a way to gather humourous material for the force's Christmas party.
Anyway, this whole operation went off fairly successfully without any offences being committed, other than various crimes against good taste by some of the more ill-dressed fairgoers. To each their own, perhaps. Naturally the police all buggered off back to their cups of tea as soon as the revellers had been ejected from the pub onto the cold damp street, at which point they all started squabbling over taxis. Well, it's not as if anyone could have predicted that....
Now, I'm not really sure why anyone would want to read about this little episode; really I'm just sticking it up here by way of introduction for my advice that you should read The Policeman's Blog, an entertaining web log written by a serving police officer. Politically it is, shall we say, a little unreconstructed (perhaps I am mistaken in taking as unironic the praise lavished upon the work of Melanie Phillips by its author) but in any case I can recommend it highly. Just as splendid is The Brick Testament, a collection of Bible stories accompanied by excruciatingly faithful (and frequently explicit) LEGO illustrations. Tip thanks to Chris Brooke and apologies that I don't have a personal anecdote to introduce the link.
3 September, 2004: But I wore the juice!
I promised more on the estimation quiz, and here it is -- better late than never, I hope. (This probably won't make much sense unless you read the first bit published on Saturday. Ignore the bit where I promised that this article would be published last Sunday.)
First a note on the scoring. Here is the distribution of scores achieved by the first five-thousand-odd respondents:
As a vague attempt to cash in on the traditional summer obsession with exam grades, I've made up some mark boundaries, too. Grading `on the curve', with 10% achieving an A, 20% a B, 40% a C, 20% a D, and the remainer failing, the grades are:
| Grade | Score (%) |
|---|---|
| fail | < 15 |
| D | < 21 |
| C | < 32 |
| B | < 42 |
| A | all others |
In the tradition of GCSEs, I could also award an A* grade for those who scored ooh, let's say, 70% or more, but I suspect most who did were cheating, so why bother?
(This explains the fairly low scores achieved by various well-informed people who answered the thing....)
I should say something about how the scores are computed, too. As I've said before, the algorithm is very ad-hoc. I wanted the scoring algorithm to:
- give an individual score for each question;
- give the maximum possible score for each question when the respondent's answer was exactly correct (including any actual error on the answer);
- give the minimum possible score for each question when the respondent's answer was claimed to be exact (uncertainty of zero) but was not correct; and
- otherwise reward accurate answers (close to the true answer) and realistic uncertainties (close to the difference between true answer and answer given), trading the two off in a reasonable fashion.
I spent ages looking for a mathematically satisfying formula for this and didn't find one. In the end I used the following for each question, where x and dx are the candidate's answer and uncertainty, and X and dX are the true answer and uncertainty: (yet again we suffer from the `HTML not being any good for maths' problem)
s² = dx² + dX²
m = x unless x = 0, in which case m = X
p = max(1 - |dx-dX| / m, 0)
score = 10 p exp( -(x-X)² / 2s²)
-- the idea here being that we give a score based on (something like) the probability of the true answer given the normal distribution implied by the given answer and uncertainty, with a penalty function which increases with the departure of the claimed uncertainty from the true error. The 10 is just there to make the score on each question a convenient integer.
There's one further hack, which is that if the question asks for a date, the answers are converted into years-before-the-present. This is so that the penalty function (which compares given uncertainties with absolute answers) is reasonable in that case, the argument being that our knowledge of the past is likely to become more uncertain the further back we go. Obviously this doesn't really work with the question on the date of birth of Christ.
On the whole, this scoring system is rubbish. Note in particular that if your uncertainty is as large as your answer, you will get zero, which hardly rewards accurate estimation of uncertainties!
Various commenters, including Andrew Cooke on Crooked Timber, and John Quiggin commenting on my last piece here, pointed out that a better approach might be to base scores on properties of the distribution of each respondents' relative errors, y = (x-X)/dx. The idea is that, if a person is good at estimating, their uncertainties should be equal to the true errors and there shouldn't be any overall bias in whether their answers are above or below the true answers (since the questions are all on different topics). In this case, y should be distributed with mean zero and variance one. This looks like a fairly reasonable way to assess scores, in fact; for instance, here are the distributions for someone who didn't do very well (score 18%):
and for someone who did rather better (42%):
One way to compare the distributions (by no means the best, but I'm lazy and it's easy to compute) is the Kolmogorov-Smirnov statistic (the largest absolute difference between the two cumulative distributions); this is pretty well correlated with the ad-hoc scores I've computed:
(Ignoring the tail that runs up to the top left, which consists mainly of people who did so well they were probably cheating, a linear model explains about 50% of the variance, which isn't too bad.)
So, next time anyone designs one of these quizzes, I'd suggest more work on the scoring algorithm along these lines. One problem with schemes based on the distribution of answers is that they can't really give a score to each answer, which could make the game a bit unsatisfying, but that's probably not a disastrous problem.
Anyway, this is a bit off-topic. The real ulterior motive of the quiz was to test the theory that people who are incompetent in a given field are also lousy at estimating their own competence. I wanted to see whether the respondents who gave the poorest answers to the estimation questions were also likely to give unreasonably narrow uncertainties.
It turns out that they don't. The uncertainties given tended to be more-or-less reasonable estimates of uncertainties. The distribution of uncertainties relative to actual errors is peaked around 1:
and looking at individual questions typically gives something like this: (in this example, for the question about the distance from the earth to the moon)
-- that is, people who gave inaccurate answers typically gave large uncertainties too. (Actually the results for some of the questions look much more complicated than that, but that's mostly because of `round number' effects, as far as I can tell.)
This probably isn't a very good test of the `incompetent and unaware of it' theory, but still, it restored my faith in humanity a bit. (I mentioned this result to a friend of mine, and he replied, ``That's disappointing.'' And then, after a pause, ``I suppose it isn't, actually.'' Which more-or-less sums it up.)
