# 'What is an XYZ item™?' Puzzles: What (Not) To Do?

What is a Duplicate Word™?

Duplicate Words™, Non-Duplicate Words™
Blueberry,Aubergine
Hill,Mountain
Manners,Polite
Hillary,Donald
Yelling,Shouting
Effort,Perseverance
Correct,Mark
Llama,Alpaca

Many people have voiced complaints about the 'What is an XYZ item™?' series kicked off by @Jlee. Some posts have properties that aren't interesting or are too obscure (shameful plug on the second one).

I'm not saying all puzzles of this type are bad, in fact the majority are great! However, for the ones that aren't as good, I'm wondering if we need some quality control.

Subjectivity also seems to be a problem in some puzzles.

What does the community think?

• Quality control is probably difficult to implement as these puzzles are just like any other puzzle. The votes(up/down) obtained, gives the idea to the OP how good the puzzle was. But yes, people are little tired of these :) And yes, they always have this mystery of how the puzzle will end up when the answer is posted :) I can think of only one rule - "It should not be anything which asks Whats on my mind? " – Techidiot Feb 11 '17 at 14:05
• Just curious, what would people think of the ones I've made? Boring? Too easy? Passable? – suomynonA Feb 23 '17 at 6:59

I admit that I'm a bit of a late-comer on the scene, having rejoined Puzzling after these sorts of puzzles started appearing, but I do believe there is a quality issue with the "What is an XYZ item™" series. The most obvious reason, of course, is that this type of puzzle is trivial to set up - pick a property, then just provide two lists of examples... right?

Unfortunately, what you want to do is usually not as simple as that, and here's my take. I've written this specifically for the Word™ series since they are the most common, but most of these tips easily apply to other variations as well.

## Do use an adequate sample size

Make sure there's enough words to work out the property. Then maybe add a few more just in case. You're the one making the puzzle, so of course the property's going to seem obvious to you, but from the perspective of the solver it'll look like there's infinite possibilities at first glance.

Aside from difficulty reasons, having more examples also makes it more likely that people will find your intended answer. At least one puzzle has fallen victim to having a much simpler unintended answer which happened to fit, and that is less than ideal.

## Don't pick examples only for the sake of having interesting definitions

A common theme I've seen is picking pairs of synonyms, one which fits the property and one which doesn't. When done poorly, this only serves to confuse and provides very little in the way of hints to the solver since the words on either side are drastically different in terms of spelling. Of course it's perfectly possible to use related word pairs to provide good clues, but make sure you actually think hard about how useful your chosen words work as clues, rather than picking whatever word pairs come to mind that happen to fit.

I prefer to think of these puzzles like playing The Witness (don't worry if you don't know what that is). The clues should guide the solver by providing adequate examples for testing and rejecting hypotheses, rather than solely for looking pretty at first glance.

## Do clue obscure properties

If the property you've chosen is obscure, that doesn't necessarily mean you shouldn't use it. What it does mean, though, is you'll likely want to provide hints.

Here are some ways hints might be provided:

• Pick a good, meaningful name for the Word™, rather than something that just tangentially relates
• Provide a bit of backstory to hint at the relevant information
• Clue a message using the words chosen
• Add a very easy example so that people have an "in" - e.g. if 31415 was present in a Number™ puzzle, that would be a good way of making puzzlers think that pi is relevant

Note that when I say hints, I mean from the get-go. Sure, you can give hints as you go if nobody gets it immediately, but ideally the puzzle should be solvable when it's initially posted.

## Don't pick arbitrary properties

In the previous point when I said obscure, I didn't mean arbitrary. Sure, you could do "Words with less than 3 consonants and with vowels in alphabetical order", but would you really like that if you were the one solving?

Think about whether you would appreciate the solution if you were the one on the solving end. Of course, opinions are subjective and there's no guarantee that everyone will like the same puzzles as you, but if you wouldn't appreciate the puzzle yourself when solving then that's a red flag.

## Do sort the words if order is irrelevant

This one's just puzzling etiquette. If the order of the words is irrelevant, sort them by alphabetical order, or by length then alphabetical order. It may seem trivial, but it gives solvers one less thing to worry about, and when the clues are out of order it's usually an indication that something is up.

• "a late-comer on the scene" ... says the guy with the 3-digit user ID :-P – Rand al'Thor Feb 11 '17 at 17:38
• Personally I really like the pairs of words that are synonyms or antonyms or some other relation. I think it's really clever in some cases and does not confuse at all because by now everyone solving these puzzles knows that these pairs are just there to be fun and surprising and haven't got anything to do with the chosen property – Ivo Beckers Feb 13 '17 at 14:20
• Frankly, I LIKE the convention of using synonyms (or not): It signals whether the meaning of a word has a bearing on the solution. – Weckar E. Feb 13 '17 at 15:10
• @IvoBeckers My opinion is that it's perfectly possible to use synonyms and still have them form excellent clues, but when people focus too hard on picking good synonyms without thinking much about solvability, that's where the problem lies (hence the emphasis on only) – Sp3000 Feb 13 '17 at 20:58
• Tried to update and clarify my thoughts on that section. – Sp3000 Feb 13 '17 at 21:12

I agree with everything in Sp3000's answer, with the proviso already discussed in its comments that there's some merit in having semantically similar examples on opposite sides to confirm that the property sought isn't a simple matter of meaning. I would add:

# Avoid subjectivity

Suppose the property you choose is something like "Words that are synonyms of names of terrible US presidents" and one of your examples is SHRUB because a shrub is like a bush and G W Bush was a terrible president. This is ... not going to be very fair on any solver who thinks G W Bush was the greatest US president of modern times, nor on anyone who thinks "bush" and "shrub" aren't synonyms.

Ideally, it should be possible for any person of reasonable intelligence to determine absolutely unambiguously whether any given word has or lacks your property.

There might be clearly-defined intermediate cases; e.g., maybe your property works by associating a number to each word and saying it must be positive; then what if it's zero? In such cases I think you should indicate somehow that something of the sort is going on. "Positive Words", "Negative Words" and "Halfway Words" or whatever.

# Provide a way in

The easiest way to make a puzzle No Fun is to make it so that the only way to solve it is to happen to think up the exact right answer, and verify that it's correct. "A Weird Word (tm) is one for which when you prepend the word 'frog' and compute its MD5 checksum, the result is a multiple of 17."

Puzzles like this for which the right solution is one of the first things you happen to think of are too easy. Puzzles for which it isn't are too difficult. A puzzle like this is very rarely just right and very rarely any fun to solve.

The "way in" might be by way of hints, as discussed in Sp3000's answer. It might be that observant solvers notice a lot of "o"s and "i"s in the Binary Words, or notice that a lot of Early Words seem to begin with letters early in the alphabet, or notice that two Totalitarian Words are anagrams of one another, or something like that.

Of course the way in may be tricky to find, or tricky to follow. But it's usually bad when there isn't one. (Possible exception: the only way to find the solution is to happen to guess it -- but the thing you have to guess is so cute that once you guess it you can't help smiling.)

(I think this is a general principle of puzzle-crafting. It is often hard to follow when making a What Is A Foo Word (tm)? puzzle, which is one reason why these puzzles are sometimes not so good.)