Consider the magic power of being able to 'reset time' in the style of Groundhog Day or The Edge of Tomorrow, ie revert one's environment and physical state to that of an earlier time while retaining memories of the previous timeline, then being able to relive or change that timeline (and being the only person who is aware that there ever was an alternative). Let's say that unlike some examples, it is not necessary to die to invoke this ability, it can be done 'at will', and that the distance back in time that a resetter jumps is not fixed, but controllable within a range that reflects their own natural ability and training.
Given that, the 'level 1' ability to jump back, say, 24 hours, is actually nowhere near as convenient or powerful as the ability to reset a shorter length of time: instead of getting wounded in a battle and having to reset the whole day and fight the whole battle again successfully as you did before until you get to the part you wanted to change, it would be more useful to be able to just jump back 10 mins (or less) and only have to re-fight the most recent stages.
This is especially true since I'm imagining butterfly effects to have a pronounced impact at making the 'new reality' diverge fairly quickly from the old one, at least on a local level. In particular, stochastic processes (like rolling dice) can't be reliably reproduced more than a few metre-seconds from the warlock's world-line.
Fighting warlocks with the ability to do 'short resets' are therefore potentially more valuable than those who can only do 'long resets', and for an ordinary leader looking to hire a warlock as a bodyguard (because they make great bodyguards), testing the range of their power is important.
It's easy for a warlock to prove that they are able to time-reset at all: you choose a random phrase and write it down, then some time later the warlock tells you what you wrote; if they get it wrong you tell them what the phrase was so they can reset and know it (although if the warlock is legitimate, you'll never see this outcome); if they get it right then they must be legit. It's also possible to test that a warlock can reset at least a certain length of time: you write down a random phrase, then a few minutes later the warlock writes something down and gives it to you (having reset into that intervening period with the future knowledge of the answer); then after the designated length of time you open both papers and compare them, showing the warlock if appropriate; the warlock needs to be able to reset back at least as far as the gap to provide the right answer.
What test can validate a warlocks' claim to be able to reset at most a certain length of time?
If a warlock is actually only able to jump back six hours but claims to be able to do one, I can't seem to devise a test that can't be fooled by the warlock taking the necessary knowledge, waiting five hours until the required 'landing time' comes within their range, then going back and pretending they jumped back one hour when they said they did. Other than methods which involve the warlock getting executed, which I'd prefer to avoid.
A few clarifications:
- There is no way for a non-warlock to detect when a warlock has reset.
- The butterfly effect from the warlock acting slightly differently after a reset can be reduced with careful training, but cannot be entirely eliminated. For instance, asking the warlock to shake a cup of dice would almost certainly change the result of the throw if the warlock reset to before the shake.
- Multiple warlocks exist (and their interaction is something I'm still grappling with), but the test should not require additional warlocks, and ideally should be immune to cheating by them (ie the candidate is not a warlock themselves, but has the assistance of someone who is).