The Revolution Will Be Variously Represented

Jeff Clark has made a word cloud of tweets from Iran:



Says Clark:

This is a Shaped Word Cloud created from the text of approximately 84,000 tweets containing the term #iranelection. The larger the word the more frequently it appears in the text. As usual you can click on a word to see the current twitter search results.

He's also maintaining a running tweet narrative that uses an algorithm to create a sort of synecdoche of tweets from Iran. Very interesting.

Unrest in Iran

Irantracker.org, from the American Enterprise Institute, is keeping track of protests and other incidents of unrest following the presidential election in Iran with this interactive map:



They also have day-by-day statistics for numbers of protesters, arrests, and deaths. They claim a total, through Thursday, of 33 deaths, 661 arrests, and more than 1.3 million protesters. They also have a lot of information and analysis on the political situation in Iran.

Via Andrew Sullivan, who's done excellent work following events in Iran and who's obsession with the situation has been feeding my own.

Red Iran, Blue Iran (False Iran, True Iran?)

We now have some supposedly "official" provincial election numbers from Iran, courtesy of Nate Silver at FiveThirtyEight, which means we can finally map the 2009 election. It was indeed - officially - a landslide:



Nate got the results on which this map is based from a student name Daniel Berman. The results are translated from Farsi, and supposedly represent the very much-disputed official results (the results that found Ahmadinejad defeating Mousavi even in Mousavi's home base of Eastern Azerbaijan). Here are the results broken down by province, with minor candidates omitted (Nate has detailed vote numbers). Provinces officially won by Mousavi are in bold:

Ardebil: Ahmadinejad 51% - Mousavi 47%
Boushehr: A 61 - M 36
Chaharmahal/Bakhtiari: A 73 - M 22
Eastern Azerbaijan: A 57 - M 42
Fars: A 70 - M 28
Ghazvin: A 73 - M 26
Ghom: A 72 - M 25
Gilan: A 68 - M 31
Golestan: A 60 - M 38
Hamedan: A 76 - M 22
Hormozgan: A 66 - M 33
Ilam: A 65 - M 31
Isfahan: A 69 - M 29
Kerman: A 78 - M 21
Kermanshah: A 59 - M 39
Khouzestan: A 65 - M 27
Kohgilouye/Boyerahmad: A 69 - M 27
Kordestan: A 53 - M 44
Lorestan: A 71 - M 23
Markazi: A 74 - M 24
Mazandaran: A 68 - M 21
Northern Khorasan: A 74 - M 25
Razavi Khorasan: A 70 - M 28
Semnan: A 78 - M 20
Sistan/Balouchestan: A 46 - M 52
Southern Khorasan: A 75 - M 24
Tehran: A 52 - M 46
Western Azerbaijan: A 47 - M 50
Yazd: A 56 - M 42
Zanjan: A 77 - M 22

This page has an Iranian province reference map, if you're curious. And Nate actually has his own map which represents the vote share on a continuous scale:



So now that we have these numbers, can we conclude that the election was a fraud? Well, as I said the other day, things are still hazy at this point and it's not really the time for definitive conclusions; but there is an awful lot of fishiness... Nate looks at the provincial votes and compares it to that 2005 election I was talking about; and he, having an actual proficiency in math and stuff, is able to do some statistical analysis of the correlations between the two votes. Here's his plot of Ahmadinejad's performance in that 2005 election against his performance in the currently disputed election (each diamond represents a province):



Says Nate:

These correlations are fairly weak, especially for the latter graph. Certainly not the kind of thing that will dissuade anyone who believes the election was tainted.

But, there are some important differences between the two races; in the first round in 2005, you had five candidates who were fairly competitive -- two conservatives, two reformists, and one (Rafsanjani) who is probably best considered a centrist (by Iranian standards). This time, you had only two candidates who received a competitive number of votes. And, obviously, Iran is a complicated and ever-changing place, with votes that may shift along ethnic fault lines in addition to political ones.

But Nate also points to a couple of specific discrepancies. In particular, conservative candidates collectively received about 20% of the vote in Lorestan in 2005, but Ahmadinejad won 71% of the vote this time around. Also, as I noted in that previous post on the portents of the 2005 election, Karroubi got more than 55% of the vote in Lorestan in 2005 (as he is an ethnic Lur and it's his home province). But in this election, with Karroubi still on the ballot, he supposedly only won 5% of the vote there. Could his vote really have cratered by 90%? And could those voters have almost universally moved to Ahmadinejad, rather than Mousavi? It scarcely seems possible, and this strikes me as one of the fishiest numbers in a whole school of them.

UPDATE: More numerical weirdnesses - one town had a turnout of 141%.

Reading the Geographical Tea Leaves in Iran

Did the conservatives steal the election in Iran? We don't know yet, but there's been more than enough fishiness to warrant asking the question. One place to look for signs that the election was stolen would be the distribution of the vote. The powers that be haven't released results by province yet, but Electoral Geography 2.0 does have this map of the Iranian presidential election of 2005:



This shows the first round of voting - again, from 2005 - in which Ahmadinejad actually came in second with 20.3% of the vote before winning the runoff against Akbar Hashemi Rafsanjani with more than 60%. In the first round Ahmadinejad won the provinces of Tehran, Qazvin, Qom, Markazi, Semnan, Esfahan, Chaharmahal and Bakhtiari, Yazd, and South Khorasan. His share of the vote varied widely; according to this table, his lowest showing was 5.6% in Sistan and Baluchestan, and his highest was 55.2% in Qom.

But apparently in this weekend's election, Ahmadinejad won rather consistently across the board. Juan Cole sees the apparently flat vote distribution as containing signs of fraud:

Top Pieces of Evidence that the Iranian Presidential Election Was Stolen

1. It is claimed that Ahmadinejad won the city of Tabriz with 57%. His main opponent, Mir Hossein Mousavi, is an Azeri from Azerbaijan province, of which Tabriz is the capital. Mousavi, according to such polls as exist in Iran and widespread anecdotal evidence, did better in cities and is popular in Azerbaijan. Certainly, his rallies there were very well attended. So for an Azeri urban center to go so heavily for Ahmadinejad just makes no sense. In past elections, Azeris voted disproportionately for even minor presidential candidates who hailed from that province.

2. Ahmadinejad is claimed to have taken Tehran by over 50%. Again, he is not popular in the cities, even, as he claims, in the poor neighborhoods, in part because his policies have produced high inflation and high unemployment. That he should have won Tehran is so unlikely as to raise real questions about these numbers. [Ahmadinejad is widely thought only to have won Tehran in 2005 because the pro-reform groups were discouraged and stayed home rather than voting.)

3. It is claimed that cleric Mehdi Karoubi, the other reformist candidate, received 320,000 votes, and that he did poorly in Iran's western provinces, even losing in Luristan. He is a Lur and is popular in the west, including in Kurdistan. Karoubi received 17 percent of the vote in the first round of presidential elections in 2005. While it is possible that his support has substantially declined since then, it is hard to believe that he would get less than one percent of the vote. Moreover, he should have at least done well in the west, which he did not...

5. Ahmadinejad's numbers were fairly standard across Iran's provinces. In past elections there have been substantial ethnic and provincial variations.

Indeed; see the map above. And some of these observations are confirmed by this BBC report, ehich says that "[t]he [vote] figures, if they are to be believed, show Mr Ahmadinejad winning strongly even in the heartland of Mr Mousavi."

Is that plausible? In the first round of 2005 voting, here is how each of candidates did in their home province, again according to this table (the home provinces are based on cursory Googlings of these candidates, so I may be wrong on a couple of them):

Ahmadinejad (Semnan province): 34.8%, first place
Karroubi (Lorestan): 55.5%, first place
Larijani (Qom): 2.3%, sixth place (though I'm not sure Qom should is really a "home"province for Larijani)
Mehralizadeh (East and West Azerbaijan; he is an ethnic Azerbaijani): 20.6% and 28.9%; first in both cases
Moin (Esfahan): 11.2%, t-fourth place
Qalibaf (Razavi Khorasan): 34.8%, first place
Rafsanjani (Kerman): 41.5%, first place

And here's how the candidates did overall:
Ahmadinejad: 20.3%
Karroubi: 18.0%
Larijani: 6.1%
Mehralizadeh: 4.6%
Moin: 14.5%
Qalibaf: 14.5%
Rafsanjani: 22.0%

The only candidates to not win their home or ethnic-base province were Moin and Larijani. Given this pattern, it would be shocking if Mousavi, a more-than-credible candidate, didn't win in his own base of support (which, like that of Mehralizadeh, is in the Azerbaijans). But apparently, according to the BBC, that is just what happened. What's more, Mousavi's broader demographic base of support is in the cities throughout Iran; so for Ahmadidinejad to have gotten 57% in Tabriz, which is one of Iran's largest cities and the capital of East Azerbaijan, as Cole reports has been reported, really seems impossible - it would be like McCain beating Obama in Chicago.

All right, all of this amounts to circumstantial evidence that something weird happened with these elections. But is it definitive? Well, again, this blog has stumbled into speculation that is a bit beyond my pay grade. Nonetheless, it's worth noting some of the things that have happened in the last few days: former President Rafsanjani has resigned from the Expediency Discernment Council and the Assembly of Experts in protest of the election results; Iranian authorities have asked foreign reporters to leave the country; Mousavi has written a letter to supporters in which he calls the election results "appalling" and "a dangerous plot"; the Election Commission is supposed to wait three days to certify the results, but they went ahead and did it right away; and, of course, there have been large and sometimes violent protests in cities across Iran. I'm no expert, but those don't strike me as the sorts of events you'd expect to see following a fairly decided election.

The Iranian Blogosphere

From Harvard (or, as the Russians say, "Garvard"), a map of Iran's online discourse:



The data represented here are part of a series in the internet and democracy project. Here's the abstract:

We used computational social network mapping in combination with human and automated content analysis to analyze the Iranian blogosphere. In contrast to the conventional wisdom that Iranian bloggers are mainly young democrats critical of the regime, we found a wide range of opinions representing religious conservative points of view as well as secular and reform-minded ones, and topics ranging from politics and human rights to poetry, religion, and pop culture. Our research indicates that the Persian blogosphere is indeed a large discussion space of approximately 60,000 routinely updated blogs featuring a rich and varied mix of bloggers. Social network analysis reveals the Iranian blogosphere to be dominated by four major network formations, or poles, with identifiable sub-clusters of bloggers within those poles. We label the poles as 1) Secular/Reformist, 2) Conservative/Religious, 3) Persian Poetry and Literature, and 4) Mixed Networks. (View the full map / view the full map in Persian.) The secular/reformist pole contains both expatriates and Iranians involved in a dialog about Iranian politics, among many other issues. The conservative/religious pole contains three distinct sub-clusters, two focused principally on religious issues and one on politics and current affairs. Given the repressive political and media environment, and high profile arrests and harassment of bloggers, one might not expect to find much political contestation in the blogosphere. However, we identified a subset of the secular/reformist pole focused intently on politics and current affairs and comprised mainly of bloggers living inside Iran, which is linked in contentious dialog with the conservative political sub-cluster. Surprisingly, a minority of bloggers in the secular/reformist pole appear to blog anonymously, even in the more politically-oriented part of it; instead, it is more common for bloggers in the religious/conservative pole to blog anonymously. Blocking of blogs by the government is less pervasive than we had assumed. Most of the blogosphere network is visible inside Iran, although the most frequently blocked blogs are clearly those in the secular/reformist pole. Given the repressive media environment in Iran today, blogs may represent the most open public communications platform for political discourse. The peer-to-peer architecture of the blogosphere is more resistant to capture or control by the state than the older, hub and spoke architecture of the mass media model.

Kind of appreciate the prominence given to poetry in the Iranian blogosphere. On behalf of the English-language blogosphere, let me just say: we really are philistines.