I play 23/18 13/7. To me the crucial issue is whether to lift the rear checker. In general, the bots love lifting the rear checker in these sorts of positions, though if you read Mike Corbett's Backgammon Problems, they sometimes do so incorrectly, so one has to be a little careful. Sometimes it's right to hang back when behind in the race and the opponent's structure is awkward, in order to increase the chances of hitting a shot. But here we are not behind in the race and the opponent's structure is hardly fragile.
Your problem reminds me of Problem 1 in Norm Wiggins's book Boards, Blots, and Double Shots, shown below.
| gnubg | 130 |
|---|---|
| |
| tchow | 158 |
Position ID: 2M0xAxDMhuVgIA Match ID: cIkUAAAAAAAA | |
In Wiggins's position, 24/18 is superior to 11/5. I was puzzled by this for a while because it seemed natural to me for Blue to get his checkers up close to attack White's back checker, plus in Wiggins's position Blue is behind in the race so one would think that Blue would want to hang back to maximize contact. However, Blue's main gameplan is to contain White's checker, and it's important to realize that attacking is not the only way to contain a single rear checker. A more patient approach, gradually strengthening one's position and in particular controlling the outfield so that the checker gets sent back as soon as it tries to escape, is often more effective.
Back to your position: 13/7 13/8 actually gives less outfield coverage if Black escapes with a 6. Lifting the blot with 23/18 has the easily overlooked advantage of creating some outfield coverage with the new spare on the 18 point. 23/18 13/7 may look risky because of the blot on the midpoint. To be sure, it is a concession. But even if it gets sent back, Yellow is hardly any worse than he is now, where he has one checker back. Yellow has a bar-point anchor, so the midpoint drops in importance. The greater outfield coverage and flexibility of 23/18 13/7 make it better than 13/8 13/7.