Short answer: total chaos.
Long answer:
No Obama or Hillary means that you're looking at a field in 2007 that shapes up like this --
Frontrunner: John Edwards ('04 VP nominee, NC Senator, no scandals at this time)
Institutional Candidates: Bill Richardson (strong ties to the Clintons, former DNC chair, strong resume), Joe Biden, Chris Dodd
Fringe lefty candidates: Dennis Kucinich, Mike Gravel
From a strategic campaign perspective, you're missing (1) an African-American candidate, and (2) a candidate running to the right of everyone else. So my guess is that someone like Al Sharpton probably runs again to meet #1, and possibly someone like Evan Bayh runs to fill the DLC slot.
----
So that's where we are in January of 2008 ITTL.
Edwards probably engages in the same strategy that got him second place and 30% of the vote in Iowa IOTL 2008: living in Iowa and running to the left.
IOTL, Obama won Iowa with 38%, and Clinton (narrowly) finished in third with 29%. Edwards was hurt late by the Des Moines Register's influential endorsement going to Obama.
ITTL, there isn't a viable left-of-center candidate for the Register to endorse, and it's hard to see them picking the conservative Bayh despite their misgivings over Edwards. So I imagine Edwards picks up a (lukewarm) endorsement which amounts to at least the 30% he won IOTL, plus an initial two-thirds of OTL's Obama voters, plus a third of OTL's Clinton voters.
Why? At the time, most Clinton supporters in Iowa were looking for the most conservative candidate, but some were liberal women who would probably go to Edwards. Similarly, most Obama voters were looking for the most viable liberal candidate, but some were just anti-Edwards. The Iowa primary electorate is very white -- Sharpton got 0% in 2004 -- so I think only giving Edwards 2/3 of the Obama vote is pretty conservative.
That gives you Edwards with something like 60-65%.
Given Iowa's viability threshhold, virtually all of the remaining voters probably go to Bayh. IOTL, Richardson got 2% and Biden 1%; that's probably something like what you'd see here. So call it: Edwards 60, Bayh 35, Richardson 2, Biden 2, Sharpton 1.
That makes it an Edwards vs. Bayh race in New Hampshire, which is a strong Clinton state but not a particularly conservative Democratic electorate. Moreover, because there's a contested Republican primary going on at the same time, you're unlikely to have a large crossover voter effect.
Edwards got 17% in NH IOTL; Obama got 36%, and Hillary won with 39%. Using the same breakdown as in Iowa gives you something like Edwards 55, Bayh 40, Richardson 5.
The race is now declared over by the press, but even if Bayh soldiers on, he's going to lose Nevada and South Carolina pretty decisively. (Bayh might win the disputed Michigan primary, which both Edwards and Obama boycotted IOTL, but I doubt he has the institutional support Clinton had to get those delegates actually seated). By January 26, Edwards will have gone 4-for-4 with more than 50% of the vote in all the contests that actually matter.
By this time IOTL, Biden, Dodd, Kucinich and Richardson had all dropped out; I see no reason to change that ITTL.
Now suppose the affair comes out early -- very early -- ITTL; sometime after the South Carolina primary but before Super Tuesday. You'd be left with a damaged front-runner (Edwards), an unpalatable challenger tagged with the 'loser' label (Bayh), and no real second-tier institutional candidate left as an alternative.
Edwards, of course, will stonewall, deny and obfuscate just as IOTL. Most of his supporters aren't going to jump ship right away.
Without Edwards' cooperation, it would be very difficult for a candidate to jump in at this point; he or she is going to miss the ballot deadline for most of the states with primaries or caucuses in March. But let's say it's Gore.
Bayh probably wins a few conservative states on Super Tuesday: Arizona, Arkansas, Missouri, Oklahoma, Tennessee, and Utah. Maybe Al Sharpton is able to eke out a narrow Louisiana. And, I suppose, without Obama's caucus-organizing strategy, you might see Bayh prevail in North Dakota and Nebraska.
With all that, Edwards is still going to rack up huge numbers of delegates in California, Massachusetts, Minnesota, New York, and New Jersey.
Here's what that means, in practical terms. By Super Tuesday, 2,129 delegates to the 2008 Democratic National Convention will have been selected on the basis of primary and caucus results. A candidate needs a majority of all delegates (including superdelegates) to win the nomination. That's 2,117.
My somewhat-more-than-back-of-the-envelope figures show that, by February 5, 2008, Edwards is going to have a minimum of 1,600 pledged, hard, hand-picked delegates to the convention. Bayh will have the remaining 500.
That leaves just 1,124 pledged delegates to be won in subsequent primaries and caucuses, plus another 850 superdelegates, plus the DNC rules committee could restore up to 313 disputed delegates from Michigan and Florida (which violated the rules); IOTL, they restored 184 of those delegates.
I don't see how those numbers gets any candidate to the magic 50.1% threshhold. If Gore gets in, gets on every ballot and runs the table -- something I am not remotely convinced he could actually do, mind you -- you're still talking about him picking up something like 800 of the remaining 1,124 pledged delegates. Even if you give Gore all 850 superdelegates, he's still going to be way short.
On the flip side: Edwards probably can't stall and hold off the impending death of his campaign long enough to get 500 more pledged delegates, and obviously he isn't going to win any superdelegates once his campaign is perceived as dead in the water.
That means Bayh gets first crack and playing kingmaker, here. The problem is that I don't think he wants the Vice Presidency. He was pretty clearly in the picture in both 2004 and 2008, and everything I've read suggests that he took himself out of the running. Also, it should go without saying that Bayh is going to be seriously pissed that the Democratic Party operatives recruited Gore to run against him rather than let him get the nomination.
So yes: if Bayh cedes his delegates to Gore, Gore can probably stop Edwards. But I don't know that Bayh is willing to do that.
Whatever happens, you're looking at a brokered convention. And maybe that's enough to send the eventual nominee down to defeat even in a Democratic wave year.