Last year Paul Weaver did a lecture on playing various 6-5s and this reminded me of it. In positions where the 8 pt. is a blot it is almost always correct to remake the eight and split 24/18. (generalizing, of course) On the other hand, when there are 3 already on the eight, the run is right. Here is that middle gray area. You have to ask yourself what factors might point you in the direction of the right play. (If I stumble across the problems I'll post them, was a nice exercise)
One of the first things I always take into account is the pip count. Here the game has just started and it's easy to realize after roll you'll be up 14 pips. That's an argument for running all the way with 24/13. Next we can look at board strength. Here you have made your 4 pt. and your opponent has failed yet to make any improvements. That means you should be in favor of any hit exchanges. Other key factors here, aside from reinforcing the 8 pt. by playing 13/8, it also unstacks the heavy midpoint, something I try to keep in mind early on in the game, development. Sticking with 24/18 also keeps your men in communication with each other, doesn't leave one 'soldier' to fend for himself in the battle.
You voted and running won out, 5 votes to 4. I'd have personally played the split and build and preferred it by more than the rollout below shows but the plays are a bit closer than I'd have thought. Readers move 24/13. I will extend this rollout to @ least 5184 games, if there is any change, I'll repost. Getting backed up on rollouts believe it or not! 11 computers w/dvd burning, music pillaging, and rollouts out the yang ... not enough apparently.
| 164 |                         |
| 161 | |
| Board image courtesy of GO-Figure | |
| # | Ply | Move | Equity | |
| 1 | R | 24/18 13/8 | 0.225 | |
| 0.9% 16.7% 55.5% 44.5% 11.7% 0.4% | ||||
| 95% confidence interval: - money cubeless eq.: 0.166 ±0.009. Rollout settings: Full rollout, 2299 games (equiv. 76968 games), played 3-ply (precise), random seed, with race database. | ||||
| 2 | R | 24/13 | 0.215 (-0.010) | |
| 0.7% 14.6% 56.0% 44.0% 11.6% 0.4% | ||||
| 95% confidence interval: - money cubeless eq.: 0.154 ±0.009. Rollout settings: Full rollout, 2299 games (equiv. 74777 games), played 3-ply (precise), random seed, with race database. | ||||