There is something fishy in the numbers Ceni is reporting for Kasai Occidental and Kinshasa: They don't match up to the numbers the voting commission gave out in the earlier releases.
For example, in Kasai Occidental the fifth partial release indicated that with 94.5% of polling stations reporting, 1,590 votes (in thousands) had been counted. On that basis, I projected that 1,667 votes would be tallied; with 2,661 registered voters, this meant a turnout of 63 percent. However, in its final numbers, Ceni says that only 1,412 votes were tallied--less than the actual number they reported for their fifth release. On that basis they claim a turnout of only 51 percent.
In Kinshasa, the fifth release indicated that 1,807 votes had been tallied out of 80.5% of stations reporting. I projected a total of 2,244 votes, out of 3,288 registered, for a turnout of 68 percent. In the end, Ceni reported 1,869 votes tallied, for a turnout of only 57 percent.
So were the numbers cut deliberately to favor Kabila?
Had my projections held, I would have expected Tshisekedi to receive 1,329 votes in Kasai Occidental and Kabila to receive 311. Ceni reports that Tshisekedi received 1,027 and Kabila received 296. This means Tshisekedi "lost" about 283 votes relative to Kabila.
In Kinshasa, I expected Tshisekedi to receive 1,437 votes, to Kabila's 542. Ceni reports Tshisekedi receiving 1,162 to Kabila's 545. Again, Tshisekedi lost votes relative to Kabila. My projections say he lost a net 278 votes.
Between these two provinces alone, Tshisekedi should have come 561,000 votes closer to Kabila than Ceni reported. That's not enough to make up the difference between them, but it's not beanbag either.