Greater wish flowchart

[Skarg]

Here's my take:

Turn 1:
Main Wizard (IQ 17+) begins casting Summon Greater Demon

Turns 2-10:
Main Wizard keeps casting Summon Greater Demon - it takes 10 turns.

Turn 7:
Another Wizard (IQ 15+, Pentagram spell) starts drawing a Pentagram to summon the demon into. (Can be the same wizard, but he'd need to do it before starting the summoning, which will require 9 more ST to maintain the pentagram during the summoning compared to another wizard waiting till turn 7 to start drawing it.)

Turn 8-10:
Apprentices try to Aid the Main Wizard's ST (so he has enough ST to summon the demon (30), and DX 15 so he'll be unlikely to fail the summoning DX roll).

Turn 10:
The wizard makes his DX roll, probably only trying if he's been aided up to adjDX 15. If he fails, the demon attacks and the pentagram caster rolls 3/IQ (which needs to be 15+ to know the spell) to contain the demon. If both rolls fail (4.63% chance for each, so combined 0.21% chance, or 1 in 466 chance), the demon attacks - hopefully you're surrounded by guards or something.

Turn 11:
The wizard attempts to compel a wish from the demon. Contest of main wizard's IQ versus the Demon's new-edition-reduced IQ of 11:

The chance of winning the contest without any auto-success/failure would be:
IQ 17: 93.9%
IQ 18: 96.4%
IQ 19: 98.0%
IQ 20: 99.0%
IQ 21: 99.5%
IQ 22: 99.8%
IQ 23: 99.9%
IQ 24: 99.98%

However, the chance of an auto-success by the demon (and not by the wizard) is 4.4%, and there's another 4.4% chance of an auto-fail by the wizard (and not by the demon), so if the GM allows those to count as failures for the wizard, then the odds of failure due to auto-results is about 8.6% combined probability, or a maximum success rate of about 91.4% - multiply by the % by IQ to get the combined odds.

That means that after about 8 tries, there's only a 48.6% chance at most of not failing at least one of them.

Sadly, the current ITL PDF now says that if this contest fails, the Demon attacks... this is a huge nerf of the risk, because the wizard can set up some defensible situation, which rather spoils any odds combinations since you can probably arrange to keep a wizard alive with adequate defensive measures set up. Which seems to me pretty unacceptible, since it invites industrial wish generation by the Wizards' Guild, unless the GM does something to change the situation, like say the demons will wise up and start not going after that wizard but start attacking the guild in clever ways instead, or something. Also, it doesn't even say it gets past the pentagram if that happens, which would mean a 95.4% chance the failure has no effect.

And of course, even if there were not that nerf, there's also the matter of the demon not blasting the wizard to ashes unless another IQ contest is won, so some annoying wizards might arrange for revivals... and if the GM lets them choose which 5 attributes to lose, they can just knock 5 off ST and then start wishing again.

A 32-point goblin can have IQ 17 or 18 to start, and could begin these antics straight away, even using starting wealth to head to the Wizards' Guild and hire some apprentices to get the summoning going immediately.

Maybe it's a good thing the demon attacks physically, as that at least gives them the demon who sees a well-defended wizard the option of teleporting away and attacking something other than the wizard that the wizard might care about and not be prepared to defend... but then the GM needs to figure out how the demon would know that and so on. Maybe he just goes on an innocent victim mayhem spree for 11 turns, telling everyone that it's that wizard's fault.

Seems to me there need to be a few more points of palpable danger in there to avoid annoyingly-safe attribute pumping.

A 32-point goblin can have IQ 17 or 18 to start, and could begin these antics straight away.

Clever demon rampage behavior seems like the least rule-bending way to make there be a real risk, and it needs the pentagram to fail somehow. Either that or have the wizard die on a loss as before (or in addition to a demon attack). Of course, I can imagine many ways to make demons more interesting and the risks of wish farming very real for my own games, but the RAW assessment looks pretty bleak (that is, too easy on farmers) to me, especially if a GM makes any easy rulings (e.g. not using auto-failures, having a failure just mean the demon snarls at the pentagram, or letting a prepared defense just defeat the demon if the wizard fails the contest).