Tuesday, October 24, 2006

Heads up

André Carrilho, caricaturist for The New York Times, does not like Andrew Sullivan. And is not subtle in saying so.

I can't believe the editors missed the allusion, so presumably they approve.

The Republicans will win

Mark my words. In the upcoming elections, the Republicans will retain control of both houses in the US, and will very likely increase their majority. They will also win the presidential elections in 2008.

Not because the people voted for them, but because the manufacturers of the voting machines did.

This is not news: researchers at Princeton, Johns Hopkins and elsewhere have highlighted problems with Diebold machines for years (as I wrote recently). But America's craven media have refused to report on it: only now are a few reluctant articles appearing in mainstream newspapers.

Diebold's CEO was on record back in 2003 saying he was committed to delivering Ohio's presidential votes to the Republicans. There's no reason to think that commitment has changed.

Yes, people do unintentionally write buggy software and make defective hardware. But to screw up on this scale requires malice. There is no mistake in what Diebold are doing. It's all going according to plan for them, except for the occasional squawks of protest from researchers at east coast universities, that Diebold figure they can safely ignore.

After all, what can the skeptical eggheads do? The best they can do is mount a legal challenge, and hope it goes all the way up to the Supreme Court -- the court that originally voted Bush in.

UPDATE: Why tamper with the machine? Just say the Democratic candidate's full name won't fit on the screen and leave out his last name (but leave in his nickname, so he's now James H. "Jim").

I'll believe that's an accidental error when it happens with Arnold Schwarzenegger.

Thursday, October 19, 2006

The SF delusion

So does God exist? Suppose you, a humble mortal, find yourself inadequate to answer that, what do you do? You go to some of the great minds of our times.

Newton, of course, believed in God. So did Einstein, though his God may have little resemblance to the figure in the Old Testament. Very well, what's good enough for them is good enough for you, you say.

(Pascal had a good argument for believing: there may or may not be a God, you may or may not believe in Him, but you can only lose if there is a God and you don't believe in Him.)

Then Richard Dawkins comes along and not only argues persuasively that there's no God, but writes a book called "The God Delusion" to explain why not. (Watch him explain it to a mock-sceptical Stephen Colbert here.)

Then the ghost of Paul Erdős comes along and says, "Well, I don't know about God, but what about the Supreme Fascist, eh? Someone's out to get me, hiding my glasses and stealing my passport."

Irrefutable?

Wednesday, October 18, 2006

Hitch and hiked deaths

Christopher Hitchens is a pretty good writer -- erudite, insightful, witty -- until the topic becomes Iraq. Then he loses it. Unfortunately, these days the topic is generally Iraq.

This week (yes, I'm late to the party) he takes aim at The Lancet for publishing a recent peer-reviewed study from Johns Hopkins on excess deaths in Iraq since the invasion, poking fun at their "imprecise" figure; he's as ignorant of statistical sampling as he is of missile technology ("Not that there are any non-ballistic missiles" -- strange that the Iraq war hawk should be ignorant of guided missiles). Similar to the side-swipe that I just made at him, he takes a swipe at a letter to the editor that Lancet published some years earlier, on excess deaths of children due to sanctions.

After that, it gets a bit murkier: I really can't understand what he's saying. The few claims he makes that I follow -- such as that the occupying forces "issue regular statistics" on civilian deaths -- seem risibly naive. But mostly he doesn't make any claims at all, only insinuations, and it's hard to tell what those are too.

Why waste time on Hitchens? Because he used to be a valuable and entertaining writer and has done useful work in the past, including exposing such people as Henry Kissinger and Mother Teresa. (I wonder why more people never asked the obvious question: how did the good Mother spend those millions of dollars she received from prizes and donors over the world, and why didn't some of that money go into improving medical facilities at her hospices?) And he's still capable of taking accurate aim, with entertaining results, at the Catholic church and other easy targets when he feels like it. But he could have been a relevant writer. If he hoped, by defecting from the left (a defection I'm quite in sympathy with), to become more in tune with the new American century, it hasn't worked. He used to be amusing; now he's a joke.

Saturday, October 14, 2006

The bad old days

My last post was on how electronic voting machines have changed India's election culture, and "booth-capturing" and "ballot-stuffing" have ceased.

Yesterday, in my city, local elections were held the old-fashioned way -- with ballot papers -- and this was the result.

Friday, October 06, 2006

Pushing the wrong buttons

Once upon a time, any election in India would be accompanied by "booth-capturing", "ballot-stuffing", and other unsavoury incidents; and the declaration of results would be accompanied by allegations from the other side of theft and fraud. Then India's elections went electronic: after extensive testing and local use for 10 years, the 2004 general elections were entirely conducted on electronic machines, and were the most peaceful and least contentious ever. The world watched and applauded.


Once upon a time, the US was the beacon of democracy. Then in 2000, the presidential election took ten days of wrangling and a Supreme Court vote to decide. Embarrassed by the international coverage of butterfly ballot papers and hanging chads, the US decided to go electronic too. But the 2004 elections were among the most contentious ever, especially in Ohio. And all signs suggest that future elections will be worse. Why?


For one thing, there's the main manufacturer of these machines, Diebold. Its CEO wrote to Republicans in 2003 that he was "committed to helping Ohio deliver its electoral votes to the president next year." (He later said he didn't mean it that way.) Subsequently there have been numerous questions raised regarding the security of these machines. Recently, Edward Felten of Princeton University and his colleagues infected them with a virus that could silently change voting results. A little earlier, they showed that a commonly-available hotel minibar key could be used to physically open a Diebold machine. And that's just in the past month. Earlier, Avi Rubin of Johns Hopkins had made several devastating critiques of e-voting security, Diebold's in particular; an archive of his observations is here.


Robert F. Kennedy, Jr, has written two recent articles in Rolling Stone, arguing strongly that the 2004 election was stolen, and that the next one will be, too.


And now it seems the US is not the only country facing problems with electronic voting. Brazilian campaigners are worried about aspects of their system, and now there are reports that Nedap voting machines used in the Netherlands and France can be programmed to steal votes too.


So why is there such a lack of controversy in India? It's all to do with the sort of voting machines in use. The above countries use special-purpose voting machines that are really full-fledged computers, with RAM, hard disks, touch screens, networking capabilities, PCMCIA, peripherals, Microsoft Windows, and yes, security holes including viruses. The Indian machines are basically adding-machines: externally they just have 16 buttons (one for each candidate, but up to 4 machines can be chained together if there are more candidates), and internally they have no operating system; all their software is hard-wired on a sealed microprocessor that cannot be rewritten or replaced without damaging the machine. Moreover, the voting machine does not store vote tallies: a separate control machine does that.


This blog posting (from 2004) goes into more details. As Slate (and others) noted at that time, sometimes a pencil is indeed better than a high-tech pen.

Monday, October 02, 2006

How many links in a foot?

Just got over with tiling the house. Plenty to say about that, and we've been saying it on email, so here I'll restrict myself to just one of my numerous pet peeves, that extends beyond this project: why, nearly 60 years after the sun set on the British empire, are we still stuck with imperial units in India?

For the most part, we do use metric units: distances are in kilometres, weights are in kilograms, and so on. But a few things, like body temperature, body height, and surveying, are still in the old units.

So we needed to buy four boxes of tiles for skirting, each containing four tiles. All tile shops price their tiles "per square foot". But these tiles are "20x20", i.e., 20 inches to a side. So how many square feet are 4 boxes of 4 tiles each? Quick now! Of course, he reached for his calculator, and after several fumbles, got an answer I had obtained with pencil and paper: 44.44 square feet.

Is that the answer? Not exactly: though the shops and the contractors are stuck in British days, the manufacturers have in fact gone metric. So these tiles are not 20 inches to a side, but 500 cm -- half a metre -- to a side. If you account for that, the true answer is about 43.05 square feet. Of course, we didn't think of that in the shop.

Now let's see how that would have worked out if the price were in square metres. How many square metres to a tile? A half squared, that is a quarter. How many tiles to a box? Four. How many square metres to a box? One. How many square metres in N boxes? N. How many pushes of calculator buttons needed to do that? Depends on whom you ask.

When will these people decide that their lives would be simpler if they used the same units the rest of the world (excepting one, but including our former imperial rulers) use -- the same units that the manufacturers use?

Or maybe it's a scam: they call the tiles 20x20, rather than a quarter square metre, to be able to overcharge by about 3%.

Well, that was actually the least of our annoyances in the exercise, but at least the job got done and the result looks fairly good. Could have been better, could have been much worse.

Oh, by the way: according to Wikipedia, there are 1 17/33 links in a foot.