that the art of getting a random sample was not well developed in 1948, and that big city political machines were also more of a factor.

How do we know that Truman's last week of campaigning was so significant in shifting voter preferences? Maybe the polls were not tapping into certain demographic groups in the right proportions.

Please link to it here.

Most of Truman's voters - rural and working class - did not have telephones. Samples skewed Republican.

This was even worse in 1936 when pollsters thought Landon would beat FDR.

Hopefully. Let's just get Obama elected.

He believed the supposed 'mood' and was completely cautious and 'safe.'  Meanwhile, Truman called him everything under the sun (good for Harry).  I also believe pollsters -- the one or two companies that did it in 1948 -- stopped well before a week was left.  Losing campaigns always trot out 1948, to go along with 'the only poll that matters is on Election Day' and 'we've got momentum.'

According to wikipedia, Gallup also assumed that undecideds would break decisively for the challenger... He supposedly learned from that experience and never did that again in future polls...

Getting ahold of people in 1948 was a lot harder than it is now... How many people actually had phones back then, or had party lines or whatever.  I read that Gallup started their business by polling via mail.  Nowadays, that is considered to be ridiculous (take a lopk at how everyone has slammed the columbus dispatch poll, even though it is favorable to us)

Let's not forget that the Chicago Tribune made the error of printing a headline before even a decent number of votes had been counted.  The paper was Republican and probably believed their endorsed candidate was a sure thing, whether those new-fangled polls were right or not.  The paper made the mistake of not looking where the votes that were in had come from, and from where the remainder of the votes would come.  

Sixty year later, the same mistakes will not be made.  It is highly unlikely that John McCain, with no incumbency, can flip enough votes to be elected POTUS.

I think it is literally impossible for the last week of campaigning to have swung enough votes to account for the discrepancy.  It's not like Truman was speaking to crowds of 20,000 undecided voters at every station stop.

Good points, but why are you slamming the Post itself? In your quote, all they are doing is repeating what the McCain campaign is saying, and then they say this:

Few analysts outside the McCain campaign appear to share this view. And pollsters this time around will not make the mistake that the Gallup organization made 60 years ago -- ending their polling more than a week before the election and missing a last-minute surge in support for Truman.

So I'd say the Post doesn't actually need to bone up on their history. Neither, probably, do the McCain pollsters, but at this point, what else can they say?

That McCain's last best hope for victory was the third debate.

If he had shown up with Osama Bin Laden trussed up and thrown over his shoulder, he could have still won.

(And while the talking-heads created a narrative in which McCain gave his best performance in that debate, the reality is that it was a gaffe-filled performance in which Obama crushed him. And the polls showing that Obama scored his widest margin of victory among undecided voters during that debate would seem to back up my impression.)

Truman won because of his program.  His economic and fiscal proposals were more radical than anything seen since.