The maps were largely accurate if you remove the shading. 08's map missed Nevada (which swung heavily toward the Democrats after the financial crisis) and North Carolina (which was decided by 14,000 votes), and Florida's tie swinging to Obama. 2012 was even better at this point, only missing Iowa (whose polling remained close until Election Day). Perhaps these early polls are worth more than the experts give them credit for.
Early polls can show trends that people from outside the state might not know. In 2008 Obama was often ahead in Indiana, a state which just doesn't ever go for a Democratic nominee for President. But I live near the Indiana line, and I then got some Indiana television... and I got the impression that Republicans were running scared. Virginia? We all knew that Virginia was solidly Republican. If it was the only former Confederate state that Jimmy Carter could not win and if Bill Clinton never won it despite seeming to be an unusually-good fit as a Democratic nominee for the state, then it was never going to go Democratic.
Polls tell us something -- like where some politicians have vulnerabilities. The common wisdom is not enough. Politicians can make more appearances in places that they think might be good for some visits. So if Indiana is within 2% of flipping Democratic, then maybe Hillary Clinton might stop by Indianapolis on the way from Albuquerque to Pittsburgh. Parties can spend resources against vulnerable opponents on the other side or shore up their own in trouble. They can also tell a politician "You're not winning there, so forget it".
Polls can tell us if campaign pitches work.
