A reader of mine emailed this week to ask if I would weigh in on the disagreement between John Judis and Michael Barone over the likelihood of a "brokered" convention. Judis and Barone wrote their respective articles when I was busy packing up dishwear, so they escaped my notice.

In case they escaped yours, here's a brief recap. Judis' piece came first. He wrote:

With former Senator Fred Thompson's entry into the presidential race, the Republicans now have at least three candidates who could have the money and votes to compete, if necessary, all the way to June 2008. And they might have to do so. Indeed, when the Republicans meet in Minneapolis-St. Paul in September 2008 to choose their nominee, they might be looking at a brokered convention.

Of course, the party has had multiple strong candidates before--in 1980, for instance, and 1988 and even in 2000. But the old schedule of primaries and caucuses was designed to winnow down the field. By March, the field was invariably reduced to two candidates, one of whom would eventually gain enough delegates through the primaries and caucuses to win the nomination. But the 2008 schedule concentrates two-thirds of the primary and caucus votes in the first month, which ends February 5. If there is no clear frontrunner by then, the primary and caucus race will probably go down to June, and perhaps to the convention.

Judis goes on to develop what seems to me to be an ad hoc way to estimate how this will come about. Nevertheless, he raises an interesting idea. It is one that I myself have hinted at several times in the past months.

Michael Barone, on the other hand, disagrees. And, in his typical fashion, he makes reference to a vital yet less-than-obvious point about why a brokered convention is unlikely.

The old-time convention was a medium through which men who seldom saw each other and often didn't know each other could communicate, negotiate and reach an agreement. And not always productively. [Snip]

Only when political operators had that information would they negotiate for real, as they did in the "smoke-filled room" in Chicago's Blackstone Hotel where Warren Harding clinched the nomination in 1920. And clever operators could transform the mood of the convention hall, as liberal Republicans did by packing the galleries with young men chanting "We Want Willkie!" in Philadelphia in 1940 and as the city's sewer commissioner did a few weeks later by piping into the loudspeakers people crying "We Want Roosevelt!" a few weeks later in Chicago.

Today the convention as a communications medium has been replaced by other media, such as long-distance telephones, frequent air travel, an abundance of public opinion polls and by the television networks' delegate counts (Martin Plissner conducted the first one for CBS in 1968; in 1976 the networks' counts held up in the very close contest between Gerald Ford and Ronald Reagan). Not to mention the Political Hotline (founded by Doug Bailey in the 1980s), the Internet, the blogosphere and Blackberries. The kind of communication that was possible only at the convention in the old days is now going on all around us.

This is why you need to read Michael Barone. No other political commentator of whom I am aware consistently offers this level of analysis.

He is exactly right. Conventions served a communications function that they no longer serve. They were once a venue for party leaders to signal preferences to one another in the absence of all the high-tech stuff we take for granted. It follows from this, as Barone argues, that brokered conventions were more likely then than they are now. Preferences were signaled at the convention, and so compromises had to be found at the convention. In today's age, preferences can be signaled earlier - and so compromises can be found before the convention.

Like I said - you need to read this guy. He's on another level altogether.

Nevertheless, I think that Barone has underestimated the likelihood of a brokered convention - though he offers a much-needed push back to Judis' overestimate.

Barone is correct to argue that preferences must be signaled before a compromise can be brokered - and because communications media have changed, preferences can be signaled earlier, and compromise can be reached before the convention. However, there has also been a change in the party system - one that could increase the chance of a brokered convention. The national conventions used to be the domain of state party leaders. This control has only been broken recently. But it has indeed been broken. The function of choosing delegates has shifted from the state parties to the voters. This has had the effect of giving candidates themselves control over delegates. Delegates who go to the convention are delegates loyal to a particular candidate, rather than a state party and its leaders.

The likelihood of a brokered convention depends not just upon the extent to which preferences can be communicated prior to the convention, but also upon how reconcilable they are. In the past, preferences were essentially reducible to regional interests. The South formed a faction, the North, the Midwest, the Mountain West, and so on. Each region's goal was to ensure that the party ticket represented the region's interests. In some circumstances, reconciling these divergent goals was problematic - but usually, it was not terribly hard to do. Candidates could be found who did not offend the interests of any side. The vice presidential spot could also be used to broker a deal.

What about a convention at which the delegates are controlled by the candidates for the top of the ticket? Compromise could be more difficult. How satisfied would the Romney faction be if Giuliani was named the top candidate and Romney was named the veep? Not very. How satisfied would the Thompson faction be? Not very. In that scenario, one faction wins and two factions lose.

This points to a potential difference between preferences of convention delegates of the past and preferences of convention delegates today. Past delegates were interested in securing regional interests. Compromise was therefore possible - "all" you'd have to do is find a ticket that can satisfy the preferences of multiple regions. Delegates today are interested in securing their respective candidates at the top of the ticket. This reduces the number of possible compromises.

In other words, preferences at this year's convention might be closer to zero-sum than in year's past. One faction's win is another faction's loss. Either Romney is at the top of the ticket, and the Romney faction wins; or he isn't, and the Romney faction loses. The same goes with the Thompson faction and the Giuliani faction. There are fewer possibilities for compromise here because the preferences of delegates are not reducible to protecting regional interests through the ticket, but actually placing specific people at the top of the ticket.

So, while technology has enhanced communications media, and therefore the possibility of reaching a compromise before the convention - the fact that the factions at the convention are "personality cults" decreases the chance that a compromise will be found.

Now, bear in mind that this scenario requires the factions at the convention to be extremely loyal to their respective candidates. Delegates will have to feel as though they have suffered a real loss if their candidate does not win the top spot. It requires a rough equality in the number of delegates between at least two candidates. It also requires that no factions will be altogether happy with the vice presidential nomination. So, I am not saying that a brokered convention is likely. There are far too many "if's" to argue that it is. What I am saying is that, though Judis has overestimated its likelihood, Barone perhaps has underestimated it. Our candidate-centered electoral politics might make compromise over presidential candidates more difficult.