Their argument that a few hundred is not a large enough sample size does not hold water. It is large enough--provided the method one uses to select the sample is accurate. If the method is not accurate, then increasing the sample size will not help. This is strictly mathematics, not really a judgement call.
The judgement call comes in determining what a valid method of selecting a sample is. While I am not sure the professional pollsters have it down for this election, I suspect they are far closer than these guys.
Trump just shot ahead in the USC poll, having come back from 4 down. That’s a 5-point swing since DNC ended.