Some time in long-ago prehistory (like maybe 1984) computers were new and the possibility of fairly rapid "milling" of statistices had arrived, a number of articles appeared, I believe in Scientific American, with a couple of "deeper" ones in American Scientist, that analyzed the "democratic election process."
The uniform conclusion of all the articles that I remember was that it simply doesn't work.
In a two-party system, both parties must be sufficiently "middle of the road" to get about half of the random voters. This leads to the possibility that one party that can "capture" a single-issue demographic group that can deliver 15% in additional votes has an absolute LOCK on an election.
The Republocrats figured out that "fundamentalist Christians" represent very nearly the magic 15%, and in addition are quite prone to "voting the way their pastor tells them to." A bonus is that this group largely doesn't vote on political issues but can be motivated to vote on religious ones, so by creating a religious issue they're largely all "new votes" undiluted by divisions created by "thinking" or other marginal, fringe, lunatic, upredictable tendencies toward "unreliable" responses at the polls.
Add in that "pulpit politics" remains exempt from restrictions placed on all other political activist and lobbying groups, and they're an extremely powerful political tool. With leaders(?) who tend toward extreme gullibility and are uncommonly susceptible to a bit of pandering, they are an immense political resource that has been quite "professionally" manipulated.
At least one of those early "computer modeling" reports commented in depth on the effect of a third candidate, with the remarkable conclusion that with three (or more) candidates of significance, the "democratic process" will always elect the candidate unsatisfactory to the majority of the people.
(A proposal was that when there are more than two candidates, each person should vote against one, and the candidate receiving the fewest NO votes should be elected. Surprisingly, at least according to the model used, this results in the most people being "somewhat satisfied," which is what most people naively believe a "democratic election" should do.)
Democrats in recent national elections have suffered from the tendency for most votes for "third party" candidates to come from persons with "liberal views," rather than from traditionally conservative voters. Do NOT THINK that Mr Nader's "reluctant" campaigns in the most recent elections were "sponsored" by any one other than the Republican National Committee.
The Democrats must get coherent and everyone with a brain needs to vote for the candidates who can win, even if in some cases it means voting for "second best." Otherwise, the Republocrats will remain in full power and we are Nazi Germany 1934.