I think your figures are wrong. It sounds like you are not taking into account players that are eliminated. While this doesn't effect first night visions (nobody is eliminated yet), it would effect the choice for the second vision target; there is little point in targeting the player that just got lynched. And then of course there is the complication of whether or not the target for the first night got lynched or not. If their previous night's target is still alive at that point, the second vision has a 1 in 2 chance of hitting. But it they used their sense the first night on the person that was lynched, then the second choice is 1 in 3.
I actually did account for this. First and foremost, I state for clarity's sake that my calculations assumed first night visions, dying visions, and living past the first lynching. The Seer's chance of success was calculated as shown below: ("CoS" == shorthand for "chance of success")
(COS on first night + (1-COS on first night)*(COS on night 1 w/ dead first night choice + COS on night 1 w/ living first night choice)
1/4 + 3/4*(1/4*1/3 + 3/4*1/2)
1/4 + 3/4*(11/24)
1/4 + 11/32
19/32
The Warlock's calculation, though I did it originally through heuristics rather than an actual calculation, looks like this:
(COS on first night + (1-COS on first night)*(COS on night 1 w/ dead first night choice + COS on night 1 w/ living first night choice and living Seer + COS w/ dead Seer (this is 0, a dead Seer cannot be searched for) )
1/4 + 3/4*(1/4*1/3 + 2/4*1/2 + 1/4*0)
1/4 + 3/4*(1/3)
1/4 + 1/4
1/2
Admittedly, there is a little bit of sloppiness in my math when it comes to the chances of each subcase of night 2 with the given heuristics, but this doesn't drastically bias the results. (*edit: I no longer think this was the case; I keep on getting myself confused on the matter...*) As stated before, these are *maximum* probabilities (all variables except choices are assumed to be beneficial to sensing roles). In the actual game, many of these assumptions will not apply, and the sensing roles will have accordingly lesser chances of success in finding their targets.
I'd do more precise calculations (and probably will later), but I need to go now. It's Game Day at my friendly local game store (FLGS).
If I have extra time later this week, I'll do some more calculations as well.