That's not quite correct - there have been large and anomalous swings in all the tracking polls, although I'm not sure if they've been this large. It may be that gallup's question or questioning methods lead to an especially volatile measure. Or they simply had a few bad days (or even *one* bad day) which poisoned their tracking poll pool until it was pushed out. Remember that one in 20 days the sample error is going to exceed the margin. We can expect it to happen at least once every three weeks.
Finally, the tracking poll is a very sensitive - some might say oversensitive - device. Some might even say questionable. My main concern is that by releasing poll results daily it creates its own feedback loop - intensified by selecting "likely voters", who are also most likely to have seen recent polls. This is just lay opinion, though.
It might be interesting to note that Zogby, the Republicans' favourite pollster, has had Bush and Gore within the poll's margin of error for the last week and probably longer (I've only been checking daily for the last little while).
To add some fuel to the fire, here are some more current polls.
Poll Gore Bush Nader Buchanan time CBS News/NYT 42 44 4 1 Oct 18-21 Wash Post/ABC 46 46 3 1 Oct 23-21 Zogby daily 42 44 5 - Oct 23-21 (Reuters/MSNBC)
My lay analysis is that it's a very close contest, has been for the last month despite the polling swings, and given the usually low voter turnout numbers, whoever shows up to vote will win.
Tautologically yours,
Marco
,--------------------------------------------------------------------------.
> | Now if you think it strange that <
> | mice should elect a government made <
> Marco Anglesio | up of cats, you just look at the <
> mpa at the-wire.com | history of Canada for the last 90 years <
> http://www.the-wire.com/~mpa | and maybe you'll see that they weren't <
> | any stupider than we are. <
> | --Tommy Douglas <
`--------------------------------------------------------------------------'