The most interesting point for me about the Derren Brown lottery thing the last week was not the 'prediction' it was the fact that he gave away the fact that he does, despite his claims to the contrary, use stooges.

The critical point was in spurious 'Wisdom of Crowds' segment.  24 members of the public closed their eyes and used 'automatic writing' to produce six numbers (or approximations thereof).  One, however (a chap with an American accent) couldn't get the writing to work, and so was called to the front to do the adding up.  When the 23 had finished, he gathered their pieces of paper, interpreted the numbers or squiggles and created six lists of 23 numbers.  He added these up, divided the totals by 23 and the results were their lottery number predictions for the second to last draw they attempted.  They allegedly got 4 numbers correct, one was off by 1 and one had the correct digits the wrong way around.  

This is interesting for a couple of reasons.  First off, Derren effectively admitted that the whole Wisdom of Crowds thing was bobbins, so how did they get the result so close.  The second is more telling - the shot showed the columns of numbers the American (for want of a better label) interpreted from the pieces of paper, and the totals they summed to.  If  you took the time to freeze frame you would have seen that the columns added to the totals, and that the totals, when divided by 23, were the numbers claimed that were so close to the actual lottery results.  
The thing is, that if 23 people pick random numbers then the chance of the summed total of these numbers being divisible by 23 and resulting in an integer (a whole number) is, unsurprisingly, pretty much 1 in 23 - or around 4%.  The problem with the 'predictions' by Derren's members of the public is that all six of the totals were exactly divisible by 23.  The odds of that happening by chance are more than 100 million to one. 

So, here are the possible scenarios:

1) The 23 people not only were able to almost perfectly predict the lottery results, they did so producing a 100 million to one set of numbers that divided neatly to integers

2) That, as they were in a closed room, the production team time shifted the showing of the draw to the people in the room, and knowing the result numbers Derren influenced them to the extent that they picked exactly the right 138 numbers to give the correct result (even though not even he claimed he was attempting to influence them, and the time window to do this would have been very tight, and with automatic writing was not producing clear numbers, rather than ones that were 'interpreted'.

3) As per point 2), the production team time-shifted, knowing the numbers required.  The American chap was a stooge who wrote down numbers they supplied him which totaled perfectly to the required results (but in their over-zealousness, the production team had them sum perfectly to values divisible by 23, were anyone to freeze-frame and check - the 100 million to one change of this happening had not occurred to them).  Conveniently, as the numbers the 23 people wrote down were via 'automatic writing' and interpreted even if they saw the lists and thought that a number they might have written was missing, then there would be a reasonable explanation.

Now you could believe that either options 1) or 2) took place, but I have no doubts that it was actually 3).