IFComp score predictions: 2018-2021

It’s your fault I entered in the first place. You were a superb cheerleader. If it places well at all, you get the credit.

Hold my beer for a sec and I’ll show you underconfident.

4 Likes

Heh, thanks!

-Wade

2 Likes

Perhaps it will be like the opening scene in Putney Swope: A corporate board must vote for the next CEO but no one may vote for himself. They elect the least likely candidate because everyone says to himself, “No one is going to vote for that guy.”

– Jim

I predict Goat Game will rank significantly higher than 27th. No reason other than that it seems more accessible than some of the games listed above it.

In particular, even though it’s a two-hour game pretty much every reviewer who’s played it will have finished one thread even if they don’t finish the entire game.

In the spreadsheet, I’m interested to see Silicon and Cells rank so closely to my own entry but not surprised since they are both standalone games with sci-fi themes.

1 Like

Not seeing any top 10 predictions for The Best Man, which completely blew me away and is a very serious candidate for entering my top-20-of-all-time. Curious! I’m awaiting the results with bated breath. (I didn’t play enough of the games to give a serious ranking of my own.)

5 Likes

It was one of a handful that just missed the cut for me. I put it in the top 10 of games I didn’t test. Part of a reason it lost my IFComp-predictive tiebreaks is, it didn’t have a Certifiable Shiny Hook, just a lot of good parts. We all say we don’t grade on shiny hooks/quick laughs/what’s popular right now, and while I think most of us minimize it, I think the bias towards something exciting and eyecatching still slips through.

That’s not to say exciting and eyecatching and funny and just plain fun should be put down, especially when it’s clearly well-executed. It’s just, I had TBM in that very upper pool and suspected people might just say “Well, entry X was more fun” which might nudge TBM out.

Just rereading it I had a lot of “neat, I missed that” moments which bode well for its long-term durability. I filed it as something that would grow on people.

I hope the author gives a postmortem as to how much they had Aiden’s character nailed down and how much they wanted to leave that up to the reader.

My goodness–I didn’t know this movie even existed, but now that I’ve read its plot and brief overviews, I very much want to see it.

4 Likes

Yeah, as Andrew said, I really liked The Best Man - I gave it the single 10 in my rankings - but thought it wouldn’t resonate as strongly with the overall pool of judges given the well-known biases of what tends to do well in the Comp. Who knows, though - especially with authors’ votes counting this year, that conventional wisdom could be wrong!

1 Like

Wow, I hope I do place 35th… Or better yet, I hope 36th place bc then I can be DEAD CENTER in the middle haha!

6 Likes

Purely based on the amount of chatter, I’d say 4x4 Archipelago and A Paradox Between Worlds are going to sweep the top two positions.

If the votes tallied total hours played on any given game, 4x4 would easily win, hands down.

2 Likes

Okay, I’ve updated the prediction spreadsheet with the actual results. The correlations are pretty high! It’s interesting to see which games over- or under-performed from the predicted results.

Interestingly, for the rank correlation with the actual ranks, the IFDB rank was better than ranking based on the linear regression score. But they’re pretty close, and all very high, at >0.9. And my “vibe check” predictions had the highest rank correlation lol.

5 Likes

Ah sadly I did not get my dream of dead center, but I GUESS 18th is pretty good too lol.

Thanks to everyone who voted btw!

10 Likes

So I got 37th place, lower than I was hoping but fair enough. But ouch, only 16 ratings total…the download-only nature of the game really harmed Cygnet Committee’s visibility.

Re: other people’s games, I’m glad that SpamZapper got in the top 10 and glad to see that Goat Game ranked a bit higher than projected (22nd instead of 27th).

(Cross-posted from another thread)

2 Likes

As a fun statistic, here are some of the most underrated games from the predicted rankings:

  1. Universal Hologram - predicted rank 41, actual rank 23
  2. Cyborg Arena - predicted rank 52, actual rank 33
  3. How it was then and how it is now - predicted rank 68, actual rank 51
  4. Finding Light - predicted rank 34, actual rank 18
  5. Brave Bear - predicted rank 60, actual rank 46
  6. Off-Season at the Dream Factory - predicted rank 19, actual rank 6
  7. We, the Remainder - predicted rank 29, actual rank 16

Edit: for comparison, the average difference between the actual and predicted ranks was 6.7.

Some other observations:

  • Parser games had an average rating of 6.15, vs 5.64 for non-parser. In fact the top 6 games were all parser; the highest-rated non-parser game was The Golden Heist (Edit: The Libonotus Cup is a parser-choice hybrid, but it was listed as parser-based on the website). This is not an universal trend throughout the recent years; I think this was just a good year for parser.
  • The relationship between length and score was somewhat more complicated; 90-minute games had the highest average score at 6.50, vs 6.27 for 2-hour games and 4.53 for 15-minute games. So longer games still generally do better (but not over 2 hours).
13 Likes

The Libonotus Cup is a hybrid of parser and choice. Brilliantly.

4 Likes

Woo woo! Underdogs!

1 Like

More random facts because I love doing data analysis for fun:

  • The average number of ratings per game is 47. This is higher than 2020 or 2019 (both at ~43) but lower than 2018 (54). This is despite IFComp having fewer entries this year than in 2018, and the fact that we allow authors to vote. I’m not sure what is happening here. Are there fewer judges overall, or are judges rating fewer games, or both? This is a worrying trend!

  • Development systems: of the systems with at least two games, Choicescript had the highest average score, at 7.19, followed by Adventuron at 6.85, Inform 7 with 6.46, Quest with 6.36, Twine at 5.79, Ink at 5.32, and Unity at 4.51. Games labeled as “Custom” had an average score of 6.2, but this doesn’t include a lot of actual custom systems (listed as Java, C, Python, etc). This is on the spreadsheet under the 2021_results table. I haven’t checked previous years, but I suspect Choicescript is consistently the highest-rating common non-parser system. Then again, it’s a sample size of 12 games total.

7 Likes

Just a quick note that the “system” entries for “Dead Account” and “Wabewalker” seem to be switched at the moment; “Dead Account” is listed as Java, “Wabewalker” as Twine, but it should be the other way round.
(Since they have the same score, it doesn’t really affect the statistical evaluation in that regard, but I thought I’d mention it.)

2 Likes

@Oddy, I’ll trade you 36th for 18th LOL.

2 Likes