20 Questions strategy ideas

[JohnH]

From watching the 20 Questions games (without participating in them myself, so far, as I've yet to catch a game in progress), it seems evident that they would go a bit smoother, and perhaps be a bit less chaotic, if there were some discussion as to what the best questions might be.

I would think that the best possible yes-no question to ask, given a complete lack of other information on the movie, is one that cuts the probability-space of the answer in half as closely as possible.

So:
- Asking the decade of a movie is a bad idea because there is too a large luck component to the answer.  If it is asked if the movie was released in the 80s, then if it was we get a lot of information, but if it wasn't we get fairly little.  Of the movies riffed so far, not counting Classics which were chosen via different requirements, the median year is 1997 (that is, about half of them came after that, and half before).   So "Was the movie released before 1997?" would make a better question than decade.  That way, we get a roughly constant amount of information regardless of what the answer might be, although as more movies are riffed that number will need adjusting.
- Most Rifftrax movies are bad, so asking a question like "Did Ebert give it more than two stars?" isn't optimal because the answer is probably no.  One-and-a-half stars might be better (since Ebert sometimes hiccups and gives a bad film two stars).  (Note that Ebert himself is not reviewing movies at the moment due to continued recovery from surgery, so that might not be a good choice depending on how long he's out.)  Anyway, in general, if the answer is likely to be either yes or no, then the question is not as good as it could be. 
- As the number of movies riffed increases, the usefulness of questions that ask if an actor has been seen in a previously-riffed movie decreases.

Anyway, that's what I've figured out about the game so far.  Any other ideas as to a useful question-asking strategy to adopt?

[shortstack]

Well, my first question is: did you read the 20 questions from 2 days ago?
If not, go now, I'll wait.

Okay.

- Most Rifftrax movies are bad, so asking a question like "Did Ebert give it more than two stars?" isn't optimal because the answer is probably no.

Not true.  Unless LOTR got less than two stars from Ebert.  Rifftrax is for both good and bad and everything in between that's a matter of taste.  Aside from that, it's probably better to ask if the movie got more that four stars or something on IMDB, since that's what the mods use.

- As the number of movies riffed increases, the usefulness of questions that ask if an actor has been seen in a previously-riffed movie decreases.

Well, only if the answer is yes.  If the answer is 'no' (as it was with the most recent 20 questions) then it narrows it down significantly.

In the last round of 20Q I was impressed with someone posing alphabetical questions- like 'does the first letter in the title (aside from a, an or the) come before M in the alphabet?'

I thought that was a good strategy.

[mrbasehart]

I think we've done pretty well so far, and that it's the things beyond our control that may have hampered us: Newbies trying to guess the movie for a free rifftrax is something that's been dealt with through the MVP award.  And as people have become more experienced with the game, the questions are becoming more disciplined and less vague. 

In other words, I think through playing the game and seeing the outcome of our results, the board has developed a natural strategy that has seen us win all but one of the 20Q's. 

[shortstack]

I think we've done pretty well so far, and that it's the things beyond our control that may have hampered us: Newbies trying to guess the movie for a free rifftrax is something that's been dealt with through the MVP award.  And as people have become more experienced with the game, the questions are becoming more disciplined and less vague. 

In other words, I think through playing the game and seeing the outcome of our results, the board has developed a natural strategy that has seen us win all but one of the 20Q's. 

I agree.  Everybody's done pretty well so far.  I think as long as people don't ask questions that are too vague, they should be creative about it. Especially with the MVP award now- the better you contribute, the more chance you'll win a free rifftrack.

HOWEVER.  the point of the game should not be to win a free track- it should be about fun! 

[MrTorso]

The worst part of the Island of Dr. Moreau 20 questions was the random quessing on top of not paying any attention to the answers that were given, and it wasn't just newbies that were at fault.

[Steve-O]

The worst part of the Island of Dr. Moreau 20 questions was the random quessing on top of not paying any attention to the answers that were given, and it wasn't just newbies that were at fault.

True dat.  But I chalk up a lot of the Dr. Moreau debacle to server instability.  The difficulty of even getting the forum to respond created a minor panic.  Between that and the lure of the free RiffTrax, some of us were posting without thinking just to be able to get a word in edgewise.

Fortunately, both issues have been fixed now, so things should run more smoothly from here on out.

[JohnH]

I've read prior 20 Questions threads, but not all of the most recent.  It's several pages before the first question shows up!  Could you summarize?

Anyway, I was just throwing out ideas.  It is possible for a 20 Questions game to be fairly hard even with good play, although the tendency of the guys to pick out better-known movies works against that.  And remember that in the Island of Dr. Moreau game, there was a bit of disbelief even when the movie was guessed correctly but hadn't been confirmed yet.

The "before or after M" question is nice for a first one, although it would take too many questions to narrow down even the first word that way. 

I agree that people who are in it for a free RiffTrax are playing for the wrong reason.  The most expensive one to date is a measly four bucks, and that's for a massive amount of riffing!  But it is fun, well for me anyway, to think about how to solve the 20 Question problem.  More fun than playing, in fact, especially when the game itself becomes a race by different people to get their question in first each turn, that's no fun at all.

[SmilinJackRoss]

The worst part of the Island of Dr. Moreau 20 questions was the random quessing on top of not paying any attention to the answers that were given, and it wasn't just newbies that were at fault.

True dat.  But I chalk up a lot of the Dr. Moreau debacle to server instability.  The difficulty of even getting the forum to respond created a minor panic.  Between that and the lure of the free RiffTrax, some of us were posting without thinking just to be able to get a word in edgewise.

Fortunately, both issues have been fixed now, so things should run more smoothly from here on out.

I really don't see us having too much trouble from here on out for these reasons. ^^^

And remember, we still figured out Dr Moreau even after wasting 6 questions last time:

-Asking "is it a romance" AFTER we knew it was sci-fi.

-After we knew it was made in the 90's, asking if it was Batttlefield Earth (2000) and Minority Report (2002).

-After we knew it was rated PG-13, asking if it was Johnny Mnemonic (R), Coneheads (PG), and 12 Monkeys (R).

I think the strategy of figuring out the genre, rating, approximate year, and even lettering in the alphabet knocks it down to something very managable.  When searching in the imdb, these pieces of information are invaluable.

[Sharktopus]

-After we knew it was rated PG-13, asking if it was Johnny Mnemonic (R), Coneheads (PG), and 12 Monkeys (R).

Coneheads?! Granted, the movie blows, but why would they riff Coneheads?

[David]

Questions that eliminate half of all remaining possible movies (whether yes or no) are definitely the best strategy. That's why I asked "after 1990?" for the opener last time (instead of merely a decade). I wonder what the best year to use for that question would be? 1997 seems like a better choice perhaps, considering how much on the recent end the spread of rifftrax films seems to be.

My alphabetical strategy can nail the first letter guaranteed in 5 questions (if you do it right), which may seem to leave a lot of movies, but if you can also nail the year in about 5 questions (assuming a quite generous 32-year span), really, how many accessible films could be left after that? And 10 questions left!

Genre seems to be a good thing to know (especially with IMDB powersearch, which I assume the mods use) but we're often wasting four or five questions just to get that. Rating is a good thing to know too, and with a much smaller number of choices for that guess.

Or you could work on the second letter, or current IMDB rating. Both easy for splitting the probability space in half. Exactly what list you have to cut in half will of course depend on what movies are left after the first wave of eliminations.

Getting the perfect strategy may be efficient, but (a) I don't see every player going along with it and (b) it may not be as "fun" to make the perfect strategy and use it every time, meaning the mods might try a different type of game. (This could be a good thing!)

Hey, let's create the ultimate 20 Questions Flow Chart to get any movie in 20 questions, guaranteed! (just kidding)

[Steve-O]

Genre seems to be a good thing to know (especially with IMDB powersearch, which I assume the mods use) but we're often wasting four or five questions just to get that. Rating is a good thing to know too, and with a much smaller number of choices for that guess.

Very good point.  The problem with genre is that, unlike title and year, genre doesn't naturally fall into a range of members that is easily halved.

Of course, we could always ask if the primary genre begins with a letter starting with M or later in the alphabet.   If it's before, we can pick from Action, Adventure, Drama, Fantasy, Horror.  If it's after, we can pick from Musical, Mystery, Romance, Sci-Fi, Thriller.

But you're also right that we may be over-thinking this.  It's just supposed to be a bit of fun, not the inspiration for a doctoral thesis.