2005-03-30

The Verdict on Kofi Annan

From an editorial in today's NYT:

The independent panel investigating possible corruption in the United Nations oil-for-food program in Iraq has delivered a mixed verdict on Secretary General Kofi Annan. On the most explosive issue, the panel largely exonerated Mr. Annan of personal corruption in the awarding of a contract to a company that employed his son.

My question is, what does it mean to be "largely exonerated"? Does this mean that Kofi Annan was or was not exonerated? To exonerate an individual is to free that individual from blame. But in my book, "largely" clearly implicates "not completely". So what exactly is Annan's status here, exonerated or not? Thus, the title of the article is misleading, as no apparent verdict is rendered.

From the article, Annan seems to be guilty of no more than an implicit complicity in the destruction of files crucial to the investigation, as well as being easily mislead by his son, Kojo, who seems to be at the heart of the issue here. But, my beef at the moment is not with the missing $21 billion from the UN's humanitarian aid program, but rather with the unnamed editorial writer, who will, perhaps, next week write an article about a woman who is "partially" pregnant.

2005-03-28

Busted by the Accents

[...] The bank attendants noted the robbers had accents, possibly Australian, and wore name tags similar to tags used by employees at the sports store Prince and Carroll worked at in Vail.

Police knew of Prince and Carroll after they were arrested in Vail in January for allegedly shooting at windows with BB guns.

When Police and the FBI matched the descriptions and accents of the January incident with the bank robbery, they circulated mug shots of Prince and Carroll.

An alert officer at Denver airport, who had earlier looked at the mug shots, identified Prince and Carroll at a security checkpoint and then, after noting the pair had accents, arrested them.

Interesting to identify robbers by accents. I wonder how effective it might be to put on a fake accent while robbing a bank?

2005-03-23

Most commonly used spam words

A list compiled by Sophos. The press release talks about the long known spammer techniques used by spammers to disguise the target words. They claim to be able to detect 5.6 billion variations of the word "Viagra", which is probably the number of possible combination according to some distance metric. But it looks a total waste of resources to be able to do it. Test even a fraction of that would be a bit dumb. Not to mention they would need to care at least about the top 10 of the list. Well, I don't have the details, but it sounds an unnecessary exaggeration. The top 5:

cialis
orgasms
viagra
shipping
milf

"Cialis" has the same function of Viagra. The meaning of "milf" is unpublishable (as half the list. What I don't buy is that debt is number 102 or credit is in the 64th position!

Now, the part which might be interesting to inguists is that this list can indicate what might be biased in a corpus harvested from the web or from discussion groups (which would be a corpus considered closer to speech).

It is also worth mentioning a recent discussion about several problems with Google counts in the Corpora List.

Sophos articles about spam: Sophos report reveals words that spammers most commonly try to disguise

2005-03-20

200 lashes for 'mocking long beards'

A SAUDI academic has been sentenced to 200 lashes and time in jail for insulting an Islamist colleague, a Saudi-owned newspaper reported today.Hamza al-Muzaini, a lecturer in linguistics at King Saud University, was accused by Abdullah al-Barak, a lecturer of Islamic culture at the same university, of defamation and insult, the London-based Al-Hayat reported.

Barak, who is described as a radical salafist - a strict form of Sunni Islam - reportedly accused Muzaini of "mocking long beards" and questioning his knowledge in an article published a few months ago, other reports said.

Muzaini was sentenced to 200 lashes, four months in prison and banned from publishing, a verdict he immediately appealed, the newspaper said.

A disturbing article that presents a harsh counterpart to the current free speech/academic speech Ward Churchill fiasco.

2005-03-09

Straight softcore

No, this post has no sexual content whatsoever. Sorry to disappoint you guys. Straight softcore was the suggestion I got from Google when I mistyped straight softare. If you seach only softare, you get the correct suggestion. But with some other word, it might become more risqué. The cause is obvious. The typo correction/suggestion system is heavily based on common searches and maybe tracks sucessive failed (or nearly failed) searches. As porn is ubiquitous and people are always looking for it, that is what you get. If one pushes a little more, one can say that Google wants you to go for porn. Of course, it is just an odd result and probably useful for some (most?) people.

But I have the impression in this case that their algorithm went too far. The spelling is not that close and straight software gives you way more hits (without quotes) than straight softcore. For me, this is one of the tradeoffs of a frequency based algorithm that we have to live with. As long as Google does not automatically corrects your typos, that should be ok.