This answer attempts to give guidance for puzzles in general, and riddles in particular where so noted.
- How does one know what is the acceptable level of "difficulty"
To know if your riddle is "acceptable" in difficulty, you first must know how difficult it is. Alas, rating your own puzzles is notoriously difficult. Your puzzle is probably harder than you think it is.
That excellent answer to What should you look for when rating the difficulty of a puzzle? discusses better than I could the challenges in trying to estimate how hard something is to solve when you already know the solution. The top answer to that question notes that the best way to know how hard your puzzle is, is to get some people to test-solve it for you.
"Difficulty" may be because the riddle (or puzzle) requires thinking about the information provided in a very specific way before it leads to a solution; or because it relies on uncommon knowledge in some specific subject or language or culture or place, or requires knowledge of some random bit of trivia; or because the provided clues are themselves mini-puzzles that have to be solved to get at their meaning. These are all ways a riddle or puzzle can be challenging, yet solvable by the right person with the right knowledge and the right way of thinking about it.
However, as happens all too often, perceived "difficulty" may also be due only to the riddle (or puzzle) being under-specified, under-clued, or poorly designed.
This includes forcing a solver to go through tedious mechanical steps instead of actual solving — please don't make your puzzle "harder" just by making it more of a grind... puzzles are, after all, supposed to be fun to solve.
Not all difficulty is equal, and not all is good.
Once you have an idea how difficult your riddle is, it's time to decide if it's too easy or too difficult to be "acceptable" here. Finding the fine line between too-easy and too-hard puzzles is a topic that comes up from time to time. The leading answer to that question suggests anything that sits unsolved on this site for more than a few hours is probably the sign of an overly difficult puzzle. I'm not sure that's always true, as some excellent puzzles have taken days or even weeks of different people thinking about different parts before someone connects the dots to come up with a solution. But it is a good general guideline, and applies pretty well to riddles, where there's not a lot of piece-wise solving; anything beyond a day or two is probably a good sign the difficulty was much higher than expected.
The flip-side of being too difficult is being too easy. Some puzzlers get frustrated that a puzzle that took many hours to craft and which they (and their test solvers?) thought was a challenge, ends up solved in mere minutes, or at least in some disappointingly small fraction of the time it took to create. If that happens to you, congratulations: you've just encountered The many-eyes effect.
The best answer I can offer, then, is:
- Get test solvers, so you eliminate the "in your own head" factor
- Try to estimate how your puzzle's difficulty compares to other puzzles here
- Make sure it's the level of difficulty you intend, and is difficult for the right reasons
- Don't be discouraged if it's solved "too quickly" - because of the many-eyes effect, solution time is not always the best indicator of difficulty
- If you really want to gauge difficulty, ask the people who solved it for their feedback
- Can it be allowed to be difficult and clues given over time, or should it be answerable within a short period of time, how do you determine how long it will take?
Actual difficulty is really up to you; I'm often amazed at what gets solved here, and often solved quickly. I already offered some rough guidance on how long a riddle should probably take to solve, but if you don't mind a riddle sitting unsolved for longer (and, potentially, answerers losing steam and then losing interest after fruitless attempts to make sense of it), you're certainly allowed to make it more inaccessible.
(It bears repeating, though, that not all difficulty is equal, and not all is good.)
You ask about time to solve. I'll reiterate that solution time is not always the best indicator of difficulty. But as it is often the case that the poster underestimates the difficulty of a puzzle and no forward progress gets made even after "long enough" has passed, hints are often a good way to give additional guidance to solvers and to make more accessible a puzzle that ended up more difficult than intended.
But here's the thing. Hints are fixes to a puzzle's difficulty. Ideally, you should not need them.
Puzzles in general should incorporate enough information in their original statement to allow someone to solve them—as part of the puzzle itself, not grafted on later as "hints". Since riddles tend to be less about solving bit by bit, and more about a global association of a single answer with all the information provided, the clues provided within the riddle need to be enough to lead to the solution.
If an added hint is effectively required for anyone not inside your head to solve the riddle then it's not a "hint", it's an essential part of the riddle--and in many cases is the only thing preventing your riddle from being "guess what I'm thinking of". That kind of information should be part of the riddle from the start.
Assuming what you're giving are actual hints, and not required information, then ...
Yes. Hints can be added after some time has passed to give extra help to solvers not making progress without them.
- How do you manage the fact that people have different problem solving skills, as well as difference in the time it takes them to solve a puzzle. As well as not having had chance to see the question before it's solved in 2 mins because it was simple
I'm not sure you do.
The audience here runs from folks cruising the Hot Network Questions list, who have only casual interest and skill in solving puzzles, to people who have a deep intuitive sense for puzzles and/or considerable knowledge and skill in solving and even creating them. It's unlikely you're going to be able to create a single puzzle that will be equally engaging to both ends of that spectrum, and equally unlikely that your puzzle will take both ends of that spectrum anything resembling similar times to solve. So you're not really going to address the disparity in people's solving skills and solving speed.
If you just don't want your puzzle to be solved in two minutes, even by experts, then don't make it too simple. A puzzle that really has just one step to solve is going to be solved as quickly as it takes someone to figure out that single step. Puzzles with multiple parts or with multiple layers are going to take a bit longer, and present an opportunity to more people to latch onto something solvable to them without them having to solve the whole puzzle in its entirety. If different parts require different knowledge, insights, and/or strategies to solve, you increase the chances that a single solver will hit a step they have to stop and think about for a while.
- Very simple puzzles are going to be solved quickly, and probably by whoever sees them first and/or types their answer fastest.
- Puzzles with multiple parts and/or multiple steps may take minutes to hours per step, frequently with different people solving different parts and posting partial solutions. Depending on the puzzle, there may or may not be a single ("meta") puzzle whose solution comes from the various parts; where there is one, it's usually one of the contributors of partial solutions who sees the big picture answer and posts it, sometimes before all the pieces have been solved.
- Complicated puzzles may take days or weeks to solve, with hints being needed along the way to guide solvers along the intended solution path. Often, different solvers will solve different steps, and someone will eventually connect the dots and find the final solution.
Different puzzle structures can lead to more people being engaged for longer periods of time and in different ways. But at the end of the day, the answer to "how do I make my puzzle last longer" is simple: More Puzzle.
This advice isn't quite so useful when we're talking about riddles, as your other questions do—riddles generally have just one layer—but for puzzles in general, creating puzzles with a solve path that goes beyond a single step can be a good way to provide solvers of any skill level a bit more challenge and engagement, and in many cases can give multiple solvers the opportunity to contribute to the solution.
- When is a riddle really "too broad"? Surely without knowing the answer to the riddle this is impossible to determine. 10 people could say on face value "this is never going to be solved" but 10 other people could instead actually solve it in 5 mins or an hour with a bit of thought
A good riddle leaves no doubt as to the correctness of its solution, once that solution is known; the answer uses all the information provided by the riddle, and while it may have to reinterpret some or all of that information, it doesn't have to make additional assumptions to justify the answer as correct.
But until a definitive answer is found, it is often the case that people will post their guesses. Poorly defined riddles tend to attract lots of guesses, as people throw anything out there to see if any of it sticks. But a lot of people attempting to answer is not what makes a puzzle "too broad".
What makes a riddle or a puzzle too broad is when people offer answers that fit with all the information provided, and where it is not apparent at all that there is a reason (beyond "that wasn't what I was thinking of") why those answers do not solve the puzzle. If there can be multiple answers for which there is no part of the puzzle that actually invalidates those answers, then those answers should be at least as valid as the solution the poster had in mind; that makes the puzzle "too broad" by definition.
Won't it often be the case that many things could fit the clues? I think this is the fun and attraction of people making answers/guesses and getting closer as a collective group. One persons answer being "close" makes others edit theirs.
If many things could fit the clues, then the puzzle is under-specified. Many things may fit some of the clues, but a good riddle will have exactly one solution that fits all the clues. Now obviously some people will go out of their way to stretch explanations to justify how their putative answer is a solution to the clues, and puzzle setters cannot close every loophole (nor should they have to), but a well-crafted puzzle will give enough information to rule in the intended solution while ruling out everything else. And it should be apparent to someone reading a wrong answer that they've either ignored parts of the puzzle, or that they've taken unwarranted liberties in fitting their "solution" to the puzzle as stated.
I appreciate your comment about the fun of watching a group get iteratively closer to a solution. Keep in mind, though, that (at least here) puzzles in general, and riddles in particular, are not interactive challenges—they're not playing Hot or Cold with the setter. If they're picking up ideas from each other about what specific clues might mean, that's fine, but those guesses should be testable by referring to the puzzle or riddle, not by needing a response from the setter as to whether they're right or not. If the puzzle lacks enough specificity to make that determination, then it's probably too broad.