Cheat.org: What's Wrong with Russia's Online Petitions?

The Russian government-sponsored initiative [ru] to increase public participation in policy-making through an online petition process [GV], was met with suspicion even before its launch [GV] in April. Now, almost two months later, opposition bloggers are crying foul, claiming that the process has already been corrupted.

Anonymous users of the Russian Public Initiative (RPI) forums [ru] first drew attention to the alleged falsifications, and the accusation was later picked up by e-Democracy specialist Leonid Volkov [ru], his friend Alexey Navalny [ru], and  major online news outlets [ru]. Specifically, forum users noticed something odd about the way votes were being accrued to a popular public initiative. With 48,000 votes, the initiative, which calls for an end to zero-tolerance Blood Alcohol Content (BAC) regulations, is currently second in popularity only to Alexey Navalny's initiative to ban bureaucrats from purchasing luxury cars. Although it has enjoyed this position since the very start of the process, bloggers claim that this popularity is at least partially contrived.

Volkov and Navalny at a 2011 rally. YouTube screenshot. May 29, 2013.

Volkov and Navalny at a 2011 rally. YouTube screenshot. May 29, 2013.

The claim was first made on the RPI forums. A guest user wrote [ru] on May 16, 2013:

Между прочим, на сайте roi.f1p.net видно накрутку голосов по инициативе с отменой нулевого промилле (по моему предположению). Там видны очень четкие ступеньки, чередующиеся с ровными участками. […] Ступеньки идут достаточно периодично, чтобы средняя скорость добавления голосов за час была без резких скачков.

By the way, on the roi.f1p.net website you can see the falsification on the initiative to repeal the zero BAC (in my opinion) [note: “in my opinion” was added later by the forum administrator]. You can see very clear steps, followed by flat portions. […] The steps are fairly periodical, so that the average speed of vote accrual per hour does not jump around.

The website the user was referring to, roi.f1p.net [ru], is a public endeavor to monitor the RPI voting process, and keeps track of the top initiatives by recording vote data every 10 minutes. It allows users to graph this data in any degree of detail they want, and indeed, if one looks closely at a 24-hour period of data for the anti-BAC initiative, it sometimes exhibits odd patterns:

A look at what the anti-BAC vote accrual looked like on May 4, 2013.

A look at what anti-BAC vote accrual looked like on May 4, 2013. From roi.f1p.net

The “steps” clearly visible above are not present in the graph of votes for Navalny's anti-luxury car measure, over the same period:

May 4, 2013 votes for luxury car ban petition.

May 4, 2013 votes for luxury car ban petition. From roi.f1p.net

What are these mysterious steps? They lose all mystery when the data are graphed on a per-10-minute basis, rather than in aggregate, which is precisely what the anonymous Svetlana S. (presumably the same anonymous guest user that first called attention to the issue on the RPI forums) did earlier this month, sending her results to Leonid Volkov, who blogged [ru] about the story on May 28:

Screenshot of a May 4 graph posted by Volkov in his blog. The blue bars are votes for Navalny's initiative, the red bars are votes for the BAC initiative. The x-axis is on a 10 minute scale.

This chart makes it clear that the “steps” in the aggregate graph from roi.f1p.net are associated with a relatively large number of “extra” votes for the BAC measure, which take place all at once, roughly once per hour. (What is less clear is why Volkov decided to omit the axis with the actual number of votes the bars represent. I downloaded the data, available here, and corroborated this and other charts posted by Volkov. On May 4, 2013, the highest BAC peak was 26 votes. In general, if vote injections did take place, they numbered 20 votes each time. The same pattern holds true for an almost two week period that roughly coincides with the so called May-holidays in Russia, i.e. from May 1 to Victory Day on May 9. During the past week the “steps” have largely disappeared from the BAC-repeal initiative, although the data continue to exhibit a much higher degree of variance than the luxury car data.)

While something clearly shady is happening with the BAC petition, the big question is why. Both Volkov and Navalny generally agree with the original whistle-blower on the RPI forums, who wrote [ru]:

Вероятно, хотят побыстрее продвинуть эту инициативу, т.к. вопрос по отмене уже решен и можно безболезненно показать, как они учитывают мнение народа (по моему предположению). А заодно уязвить Навального, который хочет, чтобы именно его инициатива первая набрала 100000 (по моему предположению).

They probably want to push this initiative along, since the question of the repeal is already decided and they can painlessly show how they are listening to the voice of the people (in my opinion). And at the same time to stick it to Navalny, who wants his initiative to be the first to make it to 100000 (in my opinion).

In particular, Volkov is of the opinion that the push for falsifying happened after President Putin signaled a willingness to address the BAC question in his direct line Q&A session on April 25, 2013. Volkov also stated that the BAC petition is catching up to the luxury car petition, in percent terms, making it seem like Navalny's initiative is in danger of being overtaken. Both Volkov and Navalny used this fact as a way to drum up more interest in voting for Navalny's initiative.

Unfortunately, these arguments are faulty. First, if there is an ongoing attempt at inflating the vote for the BAC petition, it appears to only pursue the goal of matching the luxury car initiative growth rate, rather than its overall numbers. In fact, as you can see on the graph below, the relationship between the two appears to be fairly constant from before April 25, with a gap of approximately 15 thousand votes which does not grow or shrink (Volkov's point about the gap shrinking in percentage terms is thus misleading. In this case the smallest percentage a gap of 15,000 votes can “shrink” to is 15%.) On the one hand, the relationship is almost too constant — the two curves follow each other with uncanny precision, even given the fact that some people who vote for Navalny's petition also vote for other petitions — but on the other hand there is no evidence of an attempt to somehow beat Navalny to the punch:

Graph of initiatives to ban luxury cars (orange) and repeal BAC laws (blue). From  roi.f1p.net.

Graph of votes for initiatives to ban luxury cars (orange) and repeal BAC laws (blue) trending together. From roi.f1p.net.

In response to a Twitter comment describing this issue, Volkov said [ru] that “maybe” the perpetrators are trying to keep the gap narrow in preparation for some kind of last minute push.

The other problem with Volkov's argument is also data related. It just so happens that the initiative to repeal BAC laws isn't the only one that exhibits the same strange stepwise curves we saw above. In fact, out of the top nine initiatives that the roi.f1p.net website keeps track off, all (aside from Navalny's) have the same issue. These initiatives include such disparate causes as burying Lenin:

bury lenin

May 4 data for initiative to bury Lenin.

and banning all cars besides for emergency services from using flashing sirens:

May 4 data for initiative to ban bureaucrats from using sirens on their cars.

May 4 data for initiative to ban bureaucrats from using sirens on their cars.

Assuming that roi.f1p.net data are not faulty, the fact that all of the top initiatives (except for Navalny's) share this characteristic makes it less likely that the falsifications were undertaken because of Putin's tacit approval of the BAC repeal. Of course, one could argue that the other initiatives are merely there to make the BAC repeal initiative appear to be within a general trend. This line of thinking is probably closest to the truth — the most likely scenario is that Russian Public Initiative (a government contractor), faced with a relative lack of interest in its product, is trying to fudge its own statistics to appear more successful than it actually is. Navalny's superstar initiative provides them with a baseline which they can use to simulate organic growth for other initiatives, and this is what they have been doing. The remaining question is — why so sloppy?

6 comments

Join the conversation

Authors, please log in »

Guidelines

  • All comments are reviewed by a moderator. Do not submit your comment more than once or it may be identified as spam.
  • Please treat others with respect. Comments containing hate speech, obscenity, and personal attacks will not be approved.