He actually has to be able to do so and forced movement and teleportation make this considerably easy for some monsters. In any event given this comparison is from exactly the same game, under exactly the same DM (me), with exactly the same player AND the same PC being changed from thief/executioner it isn't hard for me to tell what did more damage.How can a straight up DPR comparison work for a class that has the capacity to do automatic 2d4+1d8+stuff damage to a creature it's grabbed
Using the garrotte actually isn't even that good, the executioner in my game was using surprising charge with a rapier. That was actually really effective, but again it isn't anything that the thief couldn't do anyway. The executioner was best when dumping as MUCH stuff into a single attack as possible. Good example of a real "dump everything" approach was:vs an executioner in suboptimal conditions (the assassin doesn't have someone garroted).
Also, garrotes won't work with sneak attack.
1d8 (rapier) + 1d8 (rapier surprising charge) + 2d6 (Sneak Attack Rogue MC) + 2d10 (Assassins Strike) + 1d10 (magic item) + 1d6 (magic item) + mods + static bonus (magic item) + poison = 2d8+3d6+3d10+mods+items+poison damage. So that was a really solid nova strike and really did a lot of damage. After that his DPR was considerably unimpressive - but the initial spike damage made up for it. Even with all this, the thieves round to round DPR in any long lasting combat made it a clearly superior choice.
The other problem with the garrotte is remaining hidden from an enemy, which is sometimes absolutely impractical due to terrain, situation or enemies involved (for example, tremorsensing spiders or creatures with blindsight).