If you're going to argue that the selection of John Edwards has failed to produce a bounce, it would be helpful to provide some actual numbers from the past as a point of comparison. Since no one else has....
Until 1976, Vice Presidents were usually picked by the nominee during the week of the convention, often after the candidate had been formally nominated. Since the nomination was typically not decided until the delegates had their say, it made sense not to jump the gun. In '76, however, Ronald Reagan, trailing Gerald Ford in the delegate race, threw a Hail Mary, nominating Senator Richard Schweiker to be his running mate several weeks beforehand, then attempted to use the convention floor to force Ford to do the same. He failed, in a precursor of the nomination battle that was to follow, but the practice of picking a Veep well before the convention soon caught on.
The next time it happened was in 1984, when Walter Mondale chose Geraldine Ferraro to be his running mate a month before the convention. Selecting a woman was an unprecedented move, and such farsightedness in recognizing the existence of the "gender gap" has been fruitful for Democrats since then, but the selection caused barely a ripple in the polls. At the ensuing convention, Mondale did receive a nice bounce, but he lost anyway.
In 1988, Michael Dukakis waited until early-July to pick Lloyd Bentsen. That pick also produced a small bounce in the pre-convention polls, albeit a bounce in favor of George Bush (Dukakis picked up a much larger boost at the convention itself in late-July). When it came time for Bush to pick a nominee, his choice of Dan Quayle came in the middle of the Convention. It is safe to say that the selection of Quayle was one of the most disastrous political moves of the 20th Century, and in reaction, the American People ended up giving 41 one of the largest post-convention bounces in history.
After that, the Vice Presidential nominee was always picked well before the formal nomination process was completed, so pollsters can detect a bounce from the Veep selection as distinct from any bounce accruing from the convention itself. In each instance, the nomination of the Vice President had an almost neglible impact on the polls, while the convention itself produced a significant impact. Clinton opting for Al Gore in 1992 hardly nudged the polls at all; Gore's history-making selection of Joe Lieberman had even less of an impact. And if Dole and Bush (43) were aided by the nomination of party warhorses for the number two slot, it wasn't immediately apparent in the overnight polls.
This time, the consensus is that Kerry has picked up 2-3 points or so in the polls since he chose John Edwards (that is to say, he's gone from being tied to being about 5 points up). The White House spin machine has been quite aggressive in painting this as insignificant, and continue to act as if polls which include Ralph Nader as a "candidate" have any intellectual honesty whatsoever. In fact, a pre-convention bounce of five points, in a race in which Kerry had either trailed or been tied, is massively significant, and the state-by-state trends right now are quite favorable to the Democratic nominee. He continues to out-perform Gore in the so-called "Purple States" (states won by either Gore or Bush by <5%), and has caused a number of "Red States" to be thrown into doubt, including North Carolina, Virginia and Arizona. The cool, composed Edwards compares favorably on the stump with the vulgarian loose cannon from Wyoming, and he may have already changed the dynamic of this race.