Alternate timeline FIFP results (or: Could *any* game have beaten Counterfeit Monkey? The answer might surprise you!)

Playing
1

[EDIT: Those interested in this post might find the bracket charts at the bottom of the IFWiki summary page for the tournament useful as a visual reference. The same charts can be found on this site at https://intfiction.org/t/free-if-playoffs-rules-faq-standings/69042/6.]

While playing with data from the IFDB database, I worked out a way to compare two games via their average scores from people who have played both games. This is similar to a virtual FIFP match between them.

It’s not a perfect comparison, because the overlap of FIFP players and IFDB raters is not 100%, but in general it does a decent job of predicting winners. Of the 63 matches held, only 10 (15.9%) had results that were against prediction. (Another 5 were situations where either the computed averages were equal so there could be no prediction or the recorded votes were equal so the actual result was determined by coin toss.)

I was curious about how the tournament might have gone with only IFDB raters participating, and I found some surprising things:

In Division 1:

  • Dr Ludwig and the Devil vs. Midnight. Swordfight. in Round 1 was predicted for Ludwig by a slim margin on score, and with a bigger margin based on difference in ratings at the individual level. In reality, Midnight won.

  • Had Ludwig won in round 1, it would be predicted to lose narrowly against The Wizard Sniffer in round 2.

  • There was no prediction possible between Birdland and Of Their Shadows Deep in round 1 due to tied averages.

  • Had Birdland won in round 1, it would be predicted to narrowly win against And Then You Come to a House… in round 2.

  • Despite its real world loss, Zozzled was predicted to win against And Then You Come to a House… in round 1 by a slim margin on both average score and ratings preference.

  • Had Zozzled won against And Then You Come to a House… in round 1, then in round 2 it would be predicted to narrowly win against Of Their Shadows Deep but narrowly lose against Birdland.

  • Neither Zozzled nor Birdland would be predicted to win against The Wizard Sniffer in round 3; predictions show strong margins in both cases.

  • Interestingly, the real world result of the round 3 match of Superluminal Vagrant Twin vs. The Impossible Bottle goes against predictions. The prediction would be for a narrow win by Bottle.

  • Had The Impossible Bottle won its round 3 match, it would be predicted to lose against The Wizard Sniffer in the round 4 division championship, sending Sniffer to the semi-finals.

In Division 2:

  • There was no prediction possible between Chlorophyll and Worldsmith in round 1 due to tied averages.

  • Had Chlorophyll won against Worldsmith, there would also be no prediction for its round 2 match against According to Cain due to tied averages.

  • In round 2, The Gostak was predicted to beat Coloratura by a slim margin – in the real world, Coloratura won.

  • Against hypothetical round 3 opponents, Coloratura would be predicted to win narrowly versus either Chlorophyll or According to Cain.

  • Had The Gostak beat Coloratura in round 2, it would be predicted to prevail against either Worldsmith or According to Cain in round 3, but to narrowly lose against Chlorophyll.

  • Against other possible round 4 challengers, Counterfeit Monkey would be predicted to win versus The Gostak, Chlorophyll, or According to Cain, with Cain hypothetically presenting a surprisingly strong challenge.

In the semi-finals (round 5):

  • Had The Wizard Sniffer been Division 1 champion, the prediction would be for Counterfeit Monkey to prevail by a large margin.

Based on these projections, it seems that Counterfeit Monkey was fated to make it to the tournament championship round.

I’ll save Divisions 3 and 4 for another day.

7 Likes
2

Next, I’ll go a bit out of order and review Division 4. This division had the least agreement with hypothetical results in round 1, with half of matches going against prediction (though one of those wins was determined by coin toss). The entire top half of the division bracket seems very challenging to predict.

In Division 4:

  • Hypothetically, 4x4 Archipelago had a very narrow edge over The Impossible Stairs in round 1. In reality, it lost.

  • Per prediction, Endless, Nameless had a substantial edge over The Mulldoon Legacy in round 1. In actuality, the latter won.

  • In theory, Treasures of a Slaver’s Kingdom had a simliarly large edge over Bronze in round 1. In fact, Bronze beat it.

  • In the actual tournament, The Mulldoon Legacy was granted the win over The Impossible Stairs via coin toss in round 2. Per prediction, The Impossible Stairs has a slight edge, though the overlap of players is very small in this case so confidence of prediction is low.

  • Had Endless, Nameless won its round 1 match and gone on to face The Impossible Stairs in round 2, Stairs would be predicted to win by a small margin.

  • Had 4x4 Archipelago won its round 1 match and gone on to face The Mulldoon Legacy in round 2, Archipelago would be predicted to win by a slim margin, with low confidence due to the small overlap of players.

  • In the double counterfactual in which both 4x4 Archipelago and Endless, Nameless won their round 1 matches, no prediction is possible in their round 2 match due to tie scores.

  • Had Treasures of a Slaver’s Kingdom won its round 1 match against Bronze, predictions show uncertain results in a round 2 match against Savoir-Faire; average score and expressed preference show different winners.

  • Had Treasures of a Slaver’s Kingdom won a round 2 match against Savoir-Faire, it would be predicted to win by a small margin over The Mulldoon Legacy in round 3.

  • In the hypothetical scenarios in which Treasures of a Slaver’s Kingdom reached round 3 but was paired against something other than The Mulldoon Legacy, the predictions would be for a low-confidence loss against 4x4 Archipelago (smallest overlap yet, but strong preference shown), a win against The Impossible Stairs, and a narrow victory against Endless, Nameless.

  • In the bottom half of the bracket, the only deviation from prediction is Make It Good vs. Worlds Apart. Prediction would give a small margin to Make It Good.

  • Had Make It Good won in round 1, it would also be predicted to gain a small margin of victory over Photopia in round 2.

  • Had Make It Good made it to round 3 to face Anchorhead, the prediction would be for a narrow win by Anchorhead.

  • Had Anchorhead faced an alternate challenger in the round 4 division championship, it would be predicted to win substantially against Treasures of a Slaver’s Kingdom but only narrowly against 4x4 Archipelago.

3 Likes
3

There are reasons why some things become so canonical that they hold the throne forever-- by sheer dint of milking every drop from the form and continually being ahead of not just their own time, but any time. Shakespeare is that good. Beethoven is that good. And I can’t think of another author besides Short who is so in tune with the possibilities and beauties of the parser game. People will listen to Beethoven’s 9th forever and will always say, wow, that’s mind-blowing. People will always play Counterfeit Monkey and admire how narrative, mechanic, and world-building come together so well. I’m not sure it could be beaten unless parser games fall out of favor altogether. Or unless our true Beethoven is yet to come.

6 Likes
4

Have beaten? As in with this voting demographic in this year? No, I don’t think so. I think we could ask different populations and get different answers, but that seems like a pretty normal state of affairs.

But these things are always determined by the people voting. Who knows who those people will be in five years, or ten?

4 Likes
5

Hold those thoughts…

When first planning the Free IF Playoffs, one of the people with whom I was discussing it suggested that, even though initial matchups would be determined at random, it didn’t matter how the brackets were arranged – the final results would most likely end up reflecting the Top 100 rankings. This is, in fact, pretty much what happened; 3 of the 4 games seen in the semi-finals were the tournament’s top 3 seeds, i.e. Counterfeit Monkey, Anchorhead and Superluminal Vagrant Twin. #4 seed Worldsmith made it to the Division 2 championship match, where it was eliminated by Counterfeit Monkey, suggesting that Worldsmith would have been likely to make it to the semi-finals had it ended up in a division with no other top seeds.

I started looking at these hypothetical matchups because I was trying to answer the question: Is there any realistic scenario in which Counterfeit Monkey would not have won the tournament? I was surprised that the answer seems to be yes.

Now you’ll see why I’ve left Division 3 to be the last…

In Division 3:

  • The only round 1 match that went against prediction was Slouching Towards Bedlam vs. Will Not Let Me Go. In the virtual matchup, the latter is favored to win by a substantial margin.

  • Had Will Not Let Me Go won its round 1 match, it would also be predicted to win against The Spectators in round 2.

  • Had Will Not Let Me Go won a round 2 match, it would be predicted to lose narrowly against Repeat the Ending in round 3.

  • Although it lost in the tournament, according to prediction The Weight of a Soul is very narrowly favored over Toby’s Nose in round 2.

  • Had The Weight of a Soul won in round 2, it would be predicted to lose against Spider and Web in round 3.

  • While in real life Spider and Web won the division championship round, per prediction a win by Repeat the Ending would be expected, which would have sent it to the semi-finals.

And the most interesting results are what might have happened in the semi-finals and beyond:

  • Had Repeat the Ending won the Division 3 championship, it would be predicted to beat Anchorhead in the round 5 semi-finals. This would have sent Repeat to the tournament championship match against Counterfeit Monkey.

  • Counterfeit Monkey may have seemed destined to come out on top, but according to the hypothetical matchup methodology, the prediction for this final counterfactual match is a narrow win by Repeat the Ending.

Repeat the Ending is not the only potential victor over Counterfeit Monkey according to this methodology. Had random assignments made these matches possible, other significant possible challengers (i.e. games for which either the predicted margin or overlap of players is small enough to be relatively uncertain) include According to Cain, Spy Intrigue, Stay?, Cannery Vale, Treasures of a Slaver’s Kingdom, and Cragne Manor. Of those listed, Treasures had the best shot by the metrics, but it is still predicted to lose very narrowly.

Interestingly, Worldsmith is also predicted to be a strong challenger to Counterfeit Monkey, but the projected outcome is something of a toss-up with different hypothetical winners depending on the metric chosen. As mentioned, this matchup was actually seen in the real tournament’s Division 2 championship, a match that Monkey won handily with a score of 12-4.

Repeat the Ending is the only contestant in the entire tournament showing a prediction of a likely win against Counterfeit Monkey. As a reminder, in the comparison of hypothetical matchups and real life matches across cases where both offer clear results to compare, the winner is correctly predicted in 48 out of 58 cases. That’s about 82.8%, slightly better than four-fifths of the time.

6 Likes
6

On its face this makes sense to me - both Counterfeit Money and Repeat the Ending are high quality works that use the medium of IF to the fullest. You’d be hard pressed to adapt either of them to a traditional game, let alone any other kind of media. Every game in the playoffs is fantastic but only those two hit both the quality and uniqueness heights that I’d expect a winning game to have.

2 Likes
7

In the spirit of transparency, here are the computed metrics used for the analysis of possible alternate winners. To be counted as a “contender,” the overlap had to be at least 10 with a small difference in virtual votes and/or IFDB averages. The order of games given is by tournament seed ranking. All data is drawn from the IFDB backup for December 1, 2024.

GAME								overlap		IFDB avg CM		IFDB avg	IFDB votes CM	IFDB votes
Anchorhead							93			4.7957			4.5806		26				11
Superluminal Vagrant Twin			58			4.7931			4.5690		18				5
Worldsmith							11			4.5455			4.4545		2				3
The Wizard Sniffer					69			4.7681			4.5072		21				7
Cragne Manor						17			4.8235			4.6471		3				1
Will Not Let Me Go					30			4.8333			4.6667		10				5
Eat Me								47			4.7872			4.4468		20				5
Toby’s Nose							61			4.7869			4.4098		23				4
A Beauty Cold and Austere			34			4.8529			4.5000		13				3
Lost Pig							124			4.8065			4.4113		49				6
Savoir-Faire						55			4.8727			4.6000		15				1
The Impossible Bottle				51			4.8235			4.4314		20				3
Worlds Apart						28			4.8571			4.3929		9				2
Junior Arithmancer					28			4.7857			4.5000		12				3
Alias ‘The Magpie’					39			4.7692			4.3846		17				4
Coloratura							51			4.7255			4.5098		15				5
Stay?								14			4.7857			4.7143		2				1
The Gostak							36			4.8611			4.5833		10				1
A Long Way to the Nearest Star		21			4.8095			4.3810		8				0
Known Unknowns						28			4.8571			4.3929		12				2
Blue Lacuna							38			4.8158			4.5263		10				2
Treasures of a Slaver’s Kingdom		18			4.6667			4.6667		5				3
Violet								92			4.7935			4.3370		38				3
Spy Intrigue						16			4.8125			4.6250		2				2
Absence of Law						27			4.8519			4.2222		14				0
Spider and Web						102			4.7843			4.4314		33				8
Repeat the Ending					16			4.5000			4.6250		1				4
Birdland							54			4.8704			4.4074		20				3
Dr Ludwig and the Devil				27			4.8519			4.5385		8				1
Endless, Nameless					28			4.5714			4.2500		10				3
Cannery Vale						15			4.8000			4.6667		3				2
Make It Good						32			4.8438			4.3125		13				1
Midnight. Swordfight.				38			4.8421			4.3158		17				1
The Mulldoon Legacy					15			4.8000			4.1333		8				0
City of Secrets						34			4.8235			4.5588		11				3
Bronze								86			4.8372			4.2442		46				3
Zozzled								23			4.7826			4.2174		11				1
According to Cain					27			4.6667			4.4615		6				2
And Then You Come to a House...		32			4.8125			4.2188		16				0
Weird City Interloper				27			4.7407			4.2963		11				2
Turandot							24			4.8333			4.2917		10				1
Cryptozookeeper						9			4.4444			4.2222		2				1
4x4 Archipelago						11			4.9091			4.5455		3				0
The Axolotl Project					25			4.7600			4.1200		12				1
Foo Foo								34			4.6765			4.3235		15				4
Harmonia							29			4.9310			4.3793		14				2
Sub Rosa							31			4.8387			4.4194		11				0
Photopia							114			4.7632			4.1754		56				9
Chlorophyll							31			4.7742			4.3667		15				3
Inside the Facility					36			4.8611			4.1389		18				0
Slouching Towards Bedlam			67			4.8060			4.1642		34				3
The Impossible Stairs				19			4.7895			4.4211		8				2
Excalibur							8			4.7500			4.2857		3				0
Of Their Shadows Deep				12			4.6667			4.3333		4				1
With Those We Love Alive			39			4.8205			4.0769		21				2
Magical Makeover					22			4.7727			4.0952		12				1
The Lurking Horror II				17			4.5882			4.4118		6				3
The Weight of a Soul				16			4.6875			4.3750		6				2
The Spectators						12			4.7500			4.2500		5				1
Beautiful Dreamer					14			4.7143			4.3571		5				1
Suveh Nux							68			4.8382			4.1912		40				1
The Shadow in the Cathedral			19			4.6316			4.2632		9				2
Digital: A Love Story				7			4.4286			4.1429		4				2

Definitions:

  • overlap – number of people providing reviews for both games; note that some reviews may have no star rating (i.e. review only)
  • IFDB avg CM – the average star rating for Counterfeit Monkey given by people in overlap group
  • IFDB avg – the average star rating for the virtual challenger given by people in overlap group
  • IFDB votes CM – the number of people who provided star ratings for both games and whose star rating for Counterfeit Monkey was higher
  • IFDB votes – the number of people who provided star ratings for both games and whose star rating for the virtual challenger was higher
8

You really need to share your time travel device with the rest of us. Selfish!

5 Likes
9

The perils of typing quickly. 2024 is the right year, of course (now corrected above).

(I was tempted to make a joke about the top game of 2044 being Austen, Texas, a collaboration between you and Emily Short.)

2 Likes