If you're talking about the hundred page thread on vote flipping, then I'm sure we are talking about the same thing. The X axis was sorted by precinct from least votes to most votes, not the order in which votes came in. Logic would dictate that smaller precincts would report first, but there are plenty of reasons why that may not happen. He got all of the data well after the polls were closed...they are generally just in a .csv format that does not reflect when the votes came in, and therefore there is no way to put any measure of time into the charts. Feel free to ask RonRules if you don't' believe me, but time was not reflected in those charts.
Ah, okay. That was just starting to dawn on me as I looked more at the charts. I took "votes counted" to mean, "votes counted over time," when it really means, "votes counted in the precinct or district this data point applies to." I believe you now.

EDIT: Actually, it doesn't even mean that...
the X axis is cumulative, just not in an "over time" way, but in terms of successively summed precinct totals (presumably ordered by precinct size). No wonder I misunderstood it...
In that case, the data clearly shows lower Ron Paul support and higher Romney support in larger districts, which is what the detractors have argued all along (Romney is more popular in cities, and Ron Paul is less popular in cities). I've read proponents of the vote flipping hypothesis argue that additional data rejects the legitimacy of these demographic differences...do you know where I should look to find that again?
EDIT: Ah, I see. Historical data is used to dispute any legitimate correlation between precinct size and voter support for particular candidates, because the phenomenon is isolated to the 2012 elections, the 2008 elections, and an election in Louisiana where the guy in charge of the election just so happened to end up in jail for fraud. Any correlation at all is anomalous, whereas such a strong and pervasive correlation makes this election a statistical impossibility. Clearly, something is very different about the 2008 and 2012 elections. I suppose the alternative to fraud is that Ron Paul's unique candidacy has introduced a new and drastic correlation between candidate support and precinct size...but that explanation falls somewhat flat, since Romney's upward slope only comes at Ron Paul's expense in most states, and it comes at others' expense on occasion as well.
I'm up to speed now. It's amazing how much better I understand the whole argument now that I'm reading the charts from the right perspective...sheesh. At this point I think my biggest question is, "If this is the result of fraudulent counting, why would the caucus states appear to show this same correlation?" drummergirl suggests that if fraud is occurring, it must not be occurring so much with the local ballot counts as with the central tabulators. I'm several months behind, but the mystery thickens.