Collusion: too much collusion

In the last several months I've solo-tested Collusion a little over a dozen times and live-tested three times.  I can get through a 5p solo test in about two hours, but in live tests, we have yet to get through more than half of a game in that same amount of time.  What is happening here?

The lesson I'm learning is that there's only so quickly that certain processes can happen, and negotiation is one of those things that just can't be done quickly.  At issue is the support phase.  On your turn, you propose an action and then provide three support to actions other players have previously supported.  But you can also cut deals with people.  I had thought, and I have found in my solo tests, that in many turns, you have a general sense for which actions you like, and will tend to identify and support those very quickly, and maybe if you have a spare support or two you'll see if you can shop it around, usually in a quick little deal: support-for-a-support, that sort of thing.

What actually seems to happen is that players, when the support phase comes up, consider themselves "open for business", and survey the board, and canvass the other players, quite slowly and deliberately to see where opportunities for deals lie.  Then, once the deal-making starts, it can get deep into the weeds quickly.  "I'll give you this support disc, but I want to know what you're going to do with that, and I want a give-back clause if you change the arrangement at a later time".  

Now it must be said that players seem to really enjoy the deal making.  Every single player has said that they like the game and like the negotiation and haggling.  And every attempt I've made at adding rules to restrict the scope of deal-making has been promptly ignored by the players, usually within the very first turn of the game.  But, if each player gets (on average) about 12 turns per game, and every turn has a negotiation phase, it's easy to see how this is a recipe for extreme game length.

But what worries me even more is whether all of these individual drawn-out negotiations are necessary.  Between my turn and yours, the board hasn't changed that much; do we really need to start over, as if from scratch, and have a drawn out negotiation just a minute after we just had one, about roughly the same overall board state?  I think maybe we don't.

This makes me think that a phased structure might be better, to ensure that the negotiations are sufficiently consequential.  Thus I'm toying with replacing individual turns with a three-round 'year': in rounds 1 and 2, you propose actions in any of three baronies, and then in round 3, you provide support to actions others have proposed.  Then we evaluate all three of those baronies in one shot.

This would help deal-making in that it removes the need for "I want to know what you'll do with this disc" deal stuff; it will be much more concrete, just "I support this, you support that" -- although, I expect players will still find a way to be creative.  And we'll still provide support in order, so it's not going to become a wheeling/dealing game.  But it will mean that instead of 12-or-so negotiation phases per player per game, there will only be 5.  Could that get the length down under two hours, or will those phases balloon to occupy the vacuum that the more rigid structure leaves behind?  I'm not sure.

One concern is whether the game's rhythm still feels the same.  There's a lag between when your actions hit and when they land, there's a pacing issue where being at the front of the pack gives you freedom of choice but last in the pack gives you more control over which actions land, where in the pack do you want to sit?  This version of the game would lose some of that.  But if the payoff is a negotiation landscape that's just as satisfying but half as long, it may be a justifiable price to pay.