Not much more pointless than silly polls at this stage in the game, unless you’re trying to impact public opinion.
I disagree. They are fairly predictive, in a broad sense.
For instance, they do a good job of predicting who is going to lose. Anyone who is at 1% or less, which is a whole lot of them is not going to be the nominee. In fact quite a few of them will drop out before Iowa.
On average 23% of announced major candidates drop out before Iowa. So, in our field of 20 we can expect at least five to leave before Iowa.
That's not going to be Bernie, Biden, Warren, Harris or Buttigeig.
And, by 30 days after Iowa the average (since 1980) size of the field is 4.2. Bernie, Biden, Warren and Harris are the top 4 right now, and I'm pretty confident that they will be 4 or a small number remaining 30 days after Iowa.
Per Nate SIlver:
In addition, Biden’s bounce comes near an empirical inflection point of when early polling leads tend to hold up and when they don’t. Well-known candidates polling in the mid-30s in the early going2 are about even money to win the nomination, historically. Well-known candidates polling in the mid-to-high 20s have roughly a 1 in 4 shot, conversely.Like I said they aren't linear predictions, but they show trends and odds and suggest losers pretty well at this point.