All this talk about crowd sizes predicting election results has reminded me of an example from my own country.
The above is the final pre-election rally of
UDF in the election of 1990, the first free election in Bulgaria in 66 years. The picture doesn't really give justice to the size of the crowd - it stretched considerably further, in the direction of the upper left corner. The attendance was estimated to be up to a million - in a country that had then a population of about 9 million! Of course, under the circumstances it was easy for the UDF leadership to dismiss the polls, which showed clearly that they were losing.
Three days later UDF lost the election by 11 points.
So I'm always a bit skeptical, to put it mildly, when someone uses this argument