Fame Set And Match looks at marrying into Royalty. The Duchess of Kent got five minutes of fame, then got totally eclipsed by the others for the rest of the show. Mark Phillips got similarly ignored, as we re-trod the old, hashed, story of Di Spencer Di, and Fergie of Fergie. Sophie Rhys-Mogadon got a walk-on part, but it was all a bit cliched. Sorry.

The Observer comes up with ten crunch questions raised by the collapse of the Burrell trial. To precis...

1. It's a theft case. Why try it at the Central Criminal Courts?
2. Why was evidence from Liz Windsor admitted second hand, rather than by direct witness questionning? And why was she not summonsed in the usual way? [1]
3. The European Convention on Human Rights insists that the defence can call anyone relevant to the case. How come this is different for minor royals? [2]
4. There was mention on Friday of a Public Interest Immunity Certificate, issued by the Government to protect national security. What's the story, and who waved it about?
5. Meetings took place between the judge and prosecution from which the defence was excluded. This prejudices a fair trial. Again, what gives?
6. The prosecution alleges Burrell contributed to his own misfortune by not disclosing the meeting with Liz. If so, how come there was no award for costs against Burrell?
7. On the other hand, if this was a prosecution cock-up, how come there's no charge against the prosecution for a "wasted costs order"?
8. Ultimately, Liz's revalation is of little substance to the trial. He said he was keeping some papers, she made no comment. It's not enough to kill the case outright, given there are other documents and posessions involved. So, what's the real story behind the case being dropped?

[1] Summary of three separate questions.
[2] On the preceeding spread, there's a report on the trial of MI5 defector David Shayler, in which it's reported from court transcripts that he "has been barred" from calling Anthony Blair (MP for Sedgefield) as a witness.

In a protest against biassed refereeing in a previous game, Madagascan league runners-up Stade Olympique l'Emyrne scored 149 own goals in a match Thursday. They face the wrath of FIFA, and look set to be banned from African continental football. On the upside, the entire team has been signed by the Cincinnati Bungles, for whom a 149-0 loss represents a major achievement.

I've been thinking for some time about publishing an abstract of statistics for this site. As an experiment, here's October's results.

The trial of Paul Burrell collapsed today in a significant constitutional crisis. Burrell, you may recall, has been charged with stealing from the estate of the Diana Windsor, after her tour of Parisian nightspots came to a sudden and dramatic conclusion. After almost two years caught up somewhere in the legal hinterland, the case finally came to trial on the 14th.

Two and a half days into the trial, the jury was dismissed, and a blanket ban placed on reporting why this had occurred. Such bans tend to suggest that the jury has heard something that would prejudice them unreasonably, and that reporting the reasons would contaminate a future jury. So far, not usual, but not hugely uncommon either.

Last week, Mrs Shand-Kydd, the mother of the deceased, gave her sob story, and returned to find her own house had been burgled. Thieves with a heightened sense of irony!

The centre of the prosecution case was that Burrell had stolen the thingies concerned. "Stealing" requires an intention to deprive permanently, and a lack of permission from the original owner. Without that, it just becomes a dispute on ownership, and posession is nine tenths of the law. The prosecution's tactic was straight out of the pages of "Hurrah!" magazine - wave a wodge of aristocrats, titles, and minor royals at the jury, and hope that the citizenry will believe them when they say "the things should have belonged with Chuck / Bill / Harry because we say so" over any logical opposition. This might have worked in the 19th century, but doesn't go quite so well in this day and age.

On Tuesday, the case was unexpectedly adjourned, just before the end of the prosecution case. No reasons were given for this halt. During the break, some commentators suggested that the defence would need to call the purported owners of the items in question, Chuck Windsor (a farmer from Gloucestershire) and his son Billy (studying in Scotland.) The truth was even more bizarre.

Queen's Evidence Clears Diana Butler

...the screaming headline on Het Grauniad's webties. It emerged that Chuck's mother, Liz, had had a conversation with Burrell, and she knew he intended to keep the things. This knocked a rather large tiara-shaped hole in the prosecution case, as they couldn't now prove that lack of permission clause. The prosecution decided to stop now, rather than have the defence call their unexpected star witness, The Queen.

This is a minor constitutional crisis for very good reason. In England, charges are in practice brought by the police and a prosecution service, but the charge sheet specifies it's "Regina -v- The Accused". For the queen to give evidence for the accused means she's giving evidence against ... herself! It's conceivable that Liz's failure to disclose useful evidence could be wasting police time, but that is the sort of logical headache that the court (and The Court) can do without.

Neither police nor defence had asked HM about the meeting, mentioned in Burrell's defence notes. A decent conspiracy theorist would suggest that Madge's news only came out because the case was going to be lost. Incompetence or cover up? You deride. In the mean time, look for calls that the prosecution service should work for "The People" rather than "Regina."

The Thurthday Thearth loomth.

• 11 people searching for Virgin Trains complaints... including someone from Channel 4
• game of life board game Literally *hours* of fun for the whole family. Anya would love this one.
• st pancreas station pictures
• "are friends electric" "atomic kitten" Now *there's* a horrible thought.
• "peter hain" bastard Cabinet minister in Not A Nice Chap Shock.
• rise shite Thearcherth have the oddest way of spelling "Fame Academy".
• .ogg song avril lavigne The school bully strikes again...

"he was an oggie boy / ah said see you later boy / you're just far too free for me / yeah he was an oggie boy / open source and free for all boy / he would have cost me a phenomenal fee
now i'm with the closed source boy / all locked down and costing boy / running his fingertips down my spine / yeah i'm with the closed source boy / the fuck fair use boy / he stops people copying songs of mine."

This week's zeitgeist record is from the Foo Fighters.

Yes, those are emoticons appearing on the right. Work out for yourself what they mean. There is an internal logic, if you can find it.

A pretty annoying day, all told. Work insists that everyone takes a course on "equal opportunities and diversity", and my day is today. It's a two-day course crammed into one. Full treatment for the laws concerned, which means we're laboriously ploughing through ream upon ream of paper on the same topic, with very little room to discuss the wider issues, or the philosophy of the area.

Turns out that I've gotten something of a rep for being slightly cantankerous, questioning everything, and insisting that it is "just so". Well, there's never any room for obfuscation. Either two things are equal, or they are not. It is a very simple concept, with no fudge space. Until you start talking to social scientists, who are probably advocating Marxist dialectic theory without wanting to say so, and cleary believe every word they say before lunch, but only every third sentence after lunch.

Some stats quoted clearly don't add up: a figure on "university entrants" from "ethnic minorities" is cited as though it applies to the UK population only, but is actually all students coming to study in the UK, from any country. This sort of sloppy thinking gets the subject a deserved bad name.

From the "Completely Overtaken By Events" file: my draft coverage of Deayton for next week's Weaver's Week.

Angus Deayton looks set to lose his job presenting HAVE I GOT NEWS FOR YOU? following revalations about his private life. BBC bosses reckon Deayton, famed for his brown suits, is compromising the vaguely news-based vaguely quiz show. This assessment assumes that there is something about HIGNFY to compromise. Candidates for Mike Cable's role include regular Smart Alec Stephen Fry, and former political editor John Sergeant. BREAKFAST's new presenter Dermot Murnaghan has also been linked with the job, while radio's David Aaronovitch would be an excellent choice.

Unfortunately for my column, Deayton got his marching orders this evening.

Hidden mathematician indicator of the day. I'd love to link to a decent math blog, but I'm yet to find one, and I'm not going to compromise the quality of my work by linking to one that has three entries. This tells us something about the innate nature of mathematicians, and how they seem to prefer posting to newsgroups than constructing blogs.

Well, Formula One is going some way to cleaning up its act. Most importantly, the farce of team orders that brought the 2002 season to a crashing halt in Austria has been outlawed. Very good.

There's to be a new scoring system: 10-8-6-5-4-3-2-1 for the top eight places, which will prevent the utter domination by one team. It's not quite the top 12 places of Indycar, and there's no point (or half point) for pole or fastest lap, but it's an improvement.

Speaking of pole, the qualifying session is radically changed. Friday sees each car, in turn, race off on one flying lap. First out is the championship leader, last out the last in the standings, which will give the little guys a slight advantage. Then Saturday's session is the reverse order from Friday - slowest first, fastest last. Fastest in that session takes pole. This is long and cumbersome enough to make it worthy of at least a half point. It could also be more spectacular than the race itself.

However... this good work is completely thrown out of the window for politics. Because Belguim is taking a strong stand against tobacco advertising, its race - the best on the calendar by a country mile - has been pulled from the 2003 calendar. This is a blatantly political move by a small special interest group, as addicted to tobacco as any chain smoker, intent on subverting the democratic process of the country for its own ends. The solution is quite simple: if the Spa GP remains off the circuit, my boycott continues.

As for two circuits to chop allowing Bahrain and Red China to stage races: the utterly soulless tracks at Monza and Hungary must be favoured for a cessation of visits. I don't miss them.

Apparently, prisoners are to be called by a title, rather than just being referred to by their surname. At a stroke, this removes the one perk of being a guard: one can't shout "Archer!" at some pillock, but has to call him "Mister Archer."

This set me (and Simon Mayo, from whom comes at least one entry per week) thinking about what we like people to call us. Personally, I'm in favour of a little more formality. Anyone can call me "Mister Weaver," and I'll respond to the shorter "Weaver!" I prefer people to know me a little before they start calling me "Iain" - sales people and others who don't know me from Adam really ought to be a little more formal. I've walked away from sales before now owing to just this over-familiarity.

Apparently, the Germans introduce each other by title and full name: "Hier ist Fraulein Ulrike Labenz, und Frau Richard Karlson." Sounds like a plan!

A partial climbdown from the ramshackle renegade mob calling themselves UK Customs. Their "indicative limit" of goods that can be for personal consumption is to be raised from 800 to 3200 cigarettes. This is a guideline, representing what UK Customs deems reasonable for personal use. A 60 a day smoker will still get through the new limit in under two months, compared with barely two weeks before this change.

At the heart of the problem is, as ever, taxes. British cigarettes (and alcohol) are amongst the most highly taxed in Europe. The Finance Ministry claims the loss due to people reselling goods brought across from the continent is £3.5 billion per year; the increased revenue from a cut in duty could actually reclaim some of this amount. Reducing the absurd numbers of UK Customs officers would also contribute a positive benefit to the Treasury.

Coupled with this is the utter reluctance of the British government to accept a Europe without frontiers. Once duty has been paid on goods somewhere - anywhere! - in the European Union, there should be no barrier to its free movement within the community. This astonishingly simple idea, enshrined in the Single European Market from 1986 (European Council directive 92/12 art 8, 9 applies), is clearly still too complex for the civil servants who run the ministers who run the government.

As for the concept that people should be free to move across the continent without needing to show identity - fine, in two and a bit years time you can go from Helsinki to Lisbon via Tallinn without needing to show your passport. Unless you want to detour into the Channel Tunnel, whereupon a bunch of uniformed dunderheads will drill you in a language very similar to that used by cavedwellers. Schengen works for the rest of Europe, but spurious "security concerns", and even more spurious claims that it would be necessary for individuals to register with the authorities, have made this a complete non-issue for the UK.

The cited reasons are, of course, complete baloney. Britain is an island, but this counts for absolutely nothing in this age of plane travel and general movement. It appears to me that the civil servants, and politicians from leading parties, are running scared of the xenophobic elements in society in general, the racists who claim to be against "bogus asylum seekers" in particular, and especially the false Middle England stereotype peddled by the Daily Hell. Paul Dacre, this is your fault.

Quick link: CDs that aren't and don't work.

"I can't bear to watch any more of this," quoth a plastic pap parrot on Saturday's Pap Stars. We're right there with you, old bean. And old is the watch word - the plastician was outing himself live on national television. Not that he was gay - that doesn't matter these days. Not that he was a completely talentless performer who couldn't mime for toffee, we knew that already. No, that at twenty five years and two months, he was Too Old for this "contest." Too old to jump about and mime like a sprightly young 'un. It's a life of furry slippers and cardigans and watching COUNTDOWN for old whatever his name was.

The first person who can provide one credible reason for setting an arbitrary cut off date - *any* arbitrary cut off date, here or elsewhere in life - gets their argument put forward in full here.

On a completely separate issue, this development could result in serious repercussions for the show. This whole "drama" unfolded during the fortnightly voting session. As soon as Old Bloke told the producers that he was Old, and the producers figured they had to have him off the show, his voting lines needed to be chopped. Instantly. If they weren't, that's in blatant breach of the ICSTICS code, and Granada United Television (GRUNT) could well lose their license to print money. Er, operate premium rate phone lines.

So, we've been hit fairly badly by this Bugbear thingummy. One of its more annoying habits is to make up email names as it goes on. Along the lines of user1@domain2, where user1@domain1 is valid.

My pondering... what is the chance that A) the generated address is *valid*, or B) the generated address is *clearly wrong*. For B), we can assume that we know a correct address on sight. If you're the administrator of domain1, this is a reasonable assumption.

A) is, fairly obviously, inversely proportional to the number of addresses in your address book. Suppose you have u11@d1 u12@d1 u13@d1 u21@d2 u22@d2 Any of the three u1x usernames can be matched with d1; any of the u2x usernames matches with d2. Probability of a valid address, therefore, is
[P(u1x) * P(d1)] + [P(u2x) * P(d2)] = [.6*.6]+[.4*.4] = .36+.16 = .52

Now consider
u11@d1
u12@d1
u21@d2
u22@d2
u31@d3

Probability of a valid address? (.4)^2 + (.4)^2 + (.2)^2=.36

Note: I've assumed that a username is valid in only one domain. What if, in the above example, u11=u21? That would make P(u1x) = .6, as three of five are valid choices.
Plugging in the numbers: (.6*.4) + (.6*.4) + (.2*.2) = .24 + .24 + .04 = .52

For a larger example: ten addresses, two at each of five domains, two usernames are the same.
(.3*.2) + (.3*.2) + (.2*.2) + (.2*.2) + (.2*.2) = .06 + .06 + .04 + .04 + .04 = .24

Ten addresses at ten different domains, two usernames the same:
(.2*.1) + (.2*.1) + [.1*.1]*8 = .02 + .02 + [.01]*8 = .12

The lessons? More domains and different usernames allows for a far lower chance that an address generated at random will be valid.

I'm assuming throughout that an address that's not in the address book will be invalid. Where this is not the case, the probability of a valid address will increase.

As for question B), spotting clear fakes by inspection... I've met about five different username standards:
i) given.surname
ii) givensurname (or surnamegiven)
iii) surname.given
iv) Given name in full, surname by initial, or vice versa
v) Random characters, including (combinations of) dictionary words.

If your domain runs a type i standard, then anything without a dot (types ii, iv, v) will be clearly wrong, and type iii will probably reveal itself quickly. It's possible that a type iii address could confuse, or a type v address appear valid. A very similar argument applies for type iii addresses.

Type ii will detect i and iii by punctuation, type iv by context, and most variants of type v. A similar argument runs for type iv.

The bad news for admins who give random characters is that it's very difficult to spot the fakes by inspection. Types i and iii will appear wrong, but only if the domain doesn't use full stops. Types ii and iv will be more difficult yet. Of course, if the random characters are constrained in some way (five random alphabetic characters will cater for nearly 12 million people) this will detect many more type ii and iv entries.

If one allows users to select their own names, the chance of spotting a fake by inspection is zero, for all intents and purposes.

For the other types, the Law Of Large Numbers suggests this will fairly quickly come down to the relative usage of each type of username. For instance, if 10% of all internet addresses are of the form given.surname@domain, then up to 90% of all addresses can be rejected. (It's not quite 90% as there will be confusion with type iii addresses, and with unconstrained addresses.)

Let A be the set of people infected, and let Bugbear from person A_i draw its addresses from the contents of that person's address book. Assume that this set is A'_i, and the total collection of addresses (the complete summative union of A'_i for all i) is A'. Note that if two people have a particular address in their books, this address is twice as likely to be chosen.

A' will contain usernames from classes i through v in proportion to their use in the address books. If the set of people A is representative of the internet as a whole (say, it's drawing names from a general email list), the proportion of addresses of each class in A' will quickly tend towards their proportion in the wider internet. If, however, A is non-random - it's mostly your work colleagues, say - then addresses in A' will tend towards homogeneity, and the chances of the generated address being valid increase rapidly.

Also on the Math Geek line ... How Hard Is Tetris? Completely! (PDF link.) What's more, it's a game you're bound to lose eventually.