Big Deal or No Big Deal
March 9, 2020 9:43 AM   Subscribe

Big Deal or No Big Deal
Guess which Wikipedia articles are most popular and win big! I made a game that mashes up Wikipedia data with Deal or No Deal's mystery briefcase gameplay.

It's a fun way to find random articles and understand how Wikipedia is used. There's a little more skill involved than in Deal or No Deal, which is usually just a guessing game.

I stumbled across some fascinating stuff while making this game:
 • Silbo Gomero
 • Society for Photographing Relics of Old London
 • Cryptoconchoidsyphonostomata
 • Scho-Ka-Kola
 • Antiparallelogram
 • Mass games

Big Deal or No Big Deal is coded in React and uses the uses the Wikipedia Pageview API.
Role: designer, coder, game show host
posted by oulipian (27 comments total) 5 users marked this as a favorite


High score: $2,019,302 (2018 NFL Draft)
posted by QuakerMel at 4:00 PM on March 9, 2020 [1 favorite]

High score: $117,304 (List of constituencies of the Maharashtra Legislative Assembly)

I won my first game and I'm tempted to never try again but of course I will.
posted by General Malaise at 7:04 PM on March 9, 2020 [1 favorite]

High score: $125,439 (List of Home Improvement episodes)

Well that was a fun second try.
posted by General Malaise at 7:06 PM on March 9, 2020 [1 favorite]

Fun! I think there is a little bug - on my second game next to my high score there's a link to a nonexistent wiki page.
posted by exogenous at 10:32 AM on March 10, 2020 [1 favorite]

I've never played Deal or No Deal, so some kind of overview of the game play would be helpful. I've played twice, and I can tell that it could be fun, but I'm still not totally sure what's going on.
posted by amtho at 10:34 AM on March 10, 2020 [1 favorite]

Ah, thanks exogenous! I just fixed that bug.

amtho, I just updated the in-game messages to make it clearer what's happening. Basically, at the start you're trying to guess which article is worth the most (with pageviews translating into dollars). After that, you try to eliminate articles that you suspect are not worth as much. This increases the chance that you end up with the highest amount.

At several points during play, you're offered a "deal" that you can accept. If you're not confident that you chose a high-valued article at the start, you can accept the deal rather than risk ending up with a much lower amount.

In the gameshow Deal or No Deal, the contestants are choosing from briefcases full of money, without any clues as to what is in each case.

In my game, you can guess the relative values of the "cases" based on what you suspect is most popular on Wikipedia. I also hid the value of the highest one, to make it a little more interesting. The dollar values are based on the pageviews each article received in 2019, so the highest possible score is about 44 million. But the highest possible score in each game varies a lot, depending on what random Wikipedia articles are generated.

Fun fact: this game happened because I was teaching myself React while also watching a lot of 30 Rock. They joke about "Deal or No Deal" a lot on 30 Rock!
posted by oulipian at 11:59 AM on March 10, 2020

How is the amount that one "wins" calculated? Is it related to the difference between the initial "high-maybe" guess and the lower amounts chosen afterward?

The fact that the pageviews are for the year 2019 is relevant -- there are probably items that were really interesting to people in 2019 that would have lower relative pageviews for, say, the past 20 years.

This is cool already, just suggesting some clarifications that might add to it.
posted by amtho at 12:06 PM on March 10, 2020

Oh - also - having the different phases of the game explained when the game starts -- rather than being surprised by them as one goes along -- would help newbies :)
posted by amtho at 12:07 PM on March 10, 2020

The "deals" that you're offered during the game are simply the average of what's left on the board. (The gameshow uses a more complex formula, but I found it didn't work as well for my game.)

Thanks for the suggestions, I'm definitely gonna keep working on little improvements here and there!
posted by oulipian at 1:06 PM on March 10, 2020

So, the first thing you choose is left on the board, and everything else you choose is removed? Is there any advantage to actually choosing the highest value first?

Thanks for the explanations - they help!
posted by amtho at 7:37 PM on March 10, 2020

This is neat!

My first game had Leonardo Da Vinci and then a bunch of wildly obscure things.
So that was quite straightforward.
Later games have been more challenging.

The way Deal or No Deal works is as follows:
You choose a random box. Inside that box is a dollar value.
You then eliminate 5 other boxes. As you eliminate the boxes you see the values inside them.
You are then offered an amount to take instead of your box. So you have to decide if that amount is better than your box. So, that means that if you remove low numbers you're driving up the average expected value of your box.

If you keep going you may end up with your box, and one other unclaimed box.
So, say for arguments sake you've opened every box except the $1 and the $250,000 box.
You know that your box is one of those two boxes.
You might then be offered a choice. Your box, or $125,000.
posted by Just this guy, y'know at 3:25 AM on March 11, 2020 [1 favorite]

Is there any advantage to actually choosing the highest value first?

Yes, choosing the highest value first is ideal. That's what you're meant to try and do. You can also swap your initial choice for the last unopened case, when there is only one left.

Any deal you are offered is an average of what's left, so it will always be less than the highest one.
posted by oulipian at 8:43 AM on March 11, 2020

Like the show this is way more compelling than it has any right to be. Well done!
posted by mark k at 7:27 PM on March 13, 2020 [1 favorite]

Tried playing but it just loads forever?
posted by Rhaomi at 4:22 PM on March 19, 2020

Working now -- thanks!
posted by Rhaomi at 7:41 PM on March 19, 2020

Oh hi Rhaomi, just seeing your comments now. My website did seem a little slow yesterday, not sure why. Glad it's working!
posted by oulipian at 12:32 PM on March 20, 2020

This is really neat! I've had it open in the background for a couple days, keep forgetting about it, then coming across it and playing a few more rounds... just a few more rounds....

I'm curious, how did you end up choosing the articles that go on the board? Did you have to do anything clever, or did the distributions work out nicely enough that a really simple algorithm was effective?
posted by GSV The Structure of Our Preferred Counterfactuals at 10:42 PM on March 20, 2020

Oh there's no algorithm involved, the website simply loads 20 random Wikipedia articles (it actually requests 24, because occasionally some pages returned by the API don't have any pageviews, but only shows the first 20 with pageviews). The distribution of values just happens to work out nicely most of the time.
posted by oulipian at 12:23 PM on March 21, 2020 [1 favorite]

I wasn't able to get it to work in Chrome on Win 10 at all. Works perfectly on Firefox. Fun!
posted by kathrynm at 8:45 AM on March 30, 2020

Oh, I'm surprised it didn't work in Chrome on Win 10... I'll look into it. Thanks for the report!
posted by oulipian at 12:30 PM on March 30, 2020

This is the time suck I need these days. Love it!
posted by SisterHavana at 8:40 PM on April 4, 2020 [1 favorite]


I got lucky: "High score: $2,430,932 (Measles)"

I'm on Win10. Couldn't get it to run in Firefox (with uBlock and uMatrix, so maybe not surprising), but worked with Chrome.
posted by sarah_pdx at 5:38 PM on April 8, 2020 [1 favorite]

Ah, yeah, I've noticed that my adblocker doesn't like it – you might have to turn off uBlock for the page.
posted by oulipian at 9:22 AM on April 9, 2020

This is super fun!
posted by Sparky Buttons at 6:35 PM on April 13, 2020 [1 favorite]

As a person who used to work in Wikimedia's tech department: happy to see the creative way you are using the API!
posted by brainwane at 12:54 PM on May 20, 2020 [1 favorite]

Oh, thanks brainwane! And thanks so much for your Wikipedia work which makes things like this possible.
posted by oulipian at 9:31 AM on May 24, 2020

« Older Visualizing COVID19 with interactive programming...   |   Tracking the US Government Res... Newer »

You are not currently logged in. Log in or create a new account to post comments.