Web Paint-by-Number Forum
Topic #1021: That's Weird!
By Brian Bellis (mootpoint)

#1: Brian Bellis (mootpoint) on Dec 24, 2018

I was looking at the "View Information" window on some of my puzzles and noticed about half the people who started the puzzle never finished it. I thought that perhaps some found it difficult and just quit. Most of my puzzles are on the smaller size and typically easier but whatever.

Then I sorted my puzzles by difficulty from easiest to hardest and started looking into it. Even my very easiest puzzles that have difficulty ratings of slightly over 1 can have completion rates of only 60%.

What do you think is going on here? How can so many people not be finishing even the easiest of tiny puzzles?

Is this some sort of technicality it the way these stats are reported or are hundreds of people just not solving? I would like to hear from both experience solvers and those who open but never complete puzzles.
#2: Norma Dee (norm0908) on Dec 25, 2018
I have noticed that and have been curious. I too seem to have a completion rate around 2/3. And I always make sure my puzzles are solvable without using advanced techniques.
#3: Janez (Janez) on Dec 25, 2018
I always solve the puzzles I start. Some time, if it's too hard, I stop then I restart the next time I visit this site.
#4: Aurelian Ginkgo (AurelianGinkgo) on Dec 25, 2018
There could be various reasons why this is so, and I don't think any of them really explain the majority of incompletions. Here is a non-inclusive list of my theories. Others are welcome to add and correct.:
1. It is possible that when someone clicks into a puzzle and decides they don't want to solve it even though they haven't even started, that counts as unfinished.
2. It was too hard and they gave up.
3. It was too easy and they only want to solve challenging puzzles at the moment.
4. They noticed that the colors looked they would solve to be a mess rather than a picture (prejudging before they know either way.)
5. Someone never gets back to a puzzle that they started but intended to finish. That or it takes them a long time to return to it.
6. We also have a lot of users who have never completed even one puzzle. Perhaps they joined the site to try it out, clicked into a puzzle that happened to be yours, and discovered they didn't like solving online or pbn type logic puzzles. Then they never came back to the site.
7. They recognize the subject matter part way through and abandon the puzzle because they don't like it or are tired of it, or for whatever reason. (i.e. another Mario mushroom perfectly matching the pixels of the NES version)

The list is open.

#5: Brian Bellis (mootpoint) on Dec 25, 2018
Perhaps some might click on it to decide if they will print it out and solve on paper.

I admit to clicking on a puzzle or two and seeing a big jumble of colors and think it will be a tedious and never start solving. Do you think that is part of the unfinished count?
#6: Norma Dee (norm0908) on Dec 25, 2018
When I wasn't connected to the internet, I used to go to the library and print some to take home. I would usually click on quite a few before I made up my mind. I, too, click on puzzles just to see if I want to solve them. But it doesn't seem like these would account for the big difference. I have several with approx. 200 out of 300 finishing.
#7: Joe (infrapinklizzard) on Dec 25, 2018
In order to keep the programming simple, the numbers are probably based on how many times the puzzle has been served versus the number of correctly saved puzzles.

Okay. One brief test later:

"Players" is the number of times a puzzle has been served. If you load a puzzle page, the player count goes up. If you immediately reload it, the player count goes up again. It is not counting anything other than page loads. So it even goes up by one each time you come back to a saved puzzle since you have to load the page.

"Completed" is increased by one each time the completed flag is set in the database. Therefore, it is actually possible to have more completions than players with a little finagling. Eg.: https://www.deviantart.com/infrapinklizzard/art/Webpbn-31914-Breaking-The-System-778259947
#8: Bill Eisenmann (Bullet) on Dec 26, 2018
Two thoughts: 1. I often purposely open a puzzle, solve a little bit of it, then save it and move on to another one. I do this because I don't want to miss any! I usually solve the daily random ones from the past, and also try to keep up with the new ones. Since there is just not enough PBN time in any given day, I have a wonderful archive of puzzles. I especially like to save the huge ones, and just work a few minutes at a time on them, without neglecting my life.
2. Since I hate it when I "check" my partial solve and it offers to Undo to first error, often wiping out huge sections, I got in the habit of checking often, and saving every time. So on a medium-sized puzzle, or a difficult one, I Check and Save often. Perhaps each save counts as an incomplete solve?
Merry Christmas and Happy New Year to all!
#9: Joe (infrapinklizzard) on Dec 26, 2018
Bill, I tested that and gave the results in #6. The save itself doesn't count as a new player. If you continue on without reloading the page it also is not a new player. Only *page loads* count as +1 player.

Also: The two counts are separate.
> There are "players" (probably better thought of as "plays" which is really the number of times the page itself has been served).
> And there are "completed"s.
Any ratios or combinations of those (like "incomplete solve") is just that, a combination of the two separate stats.
#10: Brian Bellis (mootpoint) on Dec 26, 2018
So Joe. When I am making a new puzzle, I might solve and edit a puzzle many times until I am satisfied. Am I also adding to the player and completed counts or do the counts only begin after publishing?
#11: Joe (infrapinklizzard) on Dec 26, 2018
I'm pretty sure they do. I seem to remember seeing many "players" on puzzles I just published.

Having just made a test puzzle to test this, I can confirm that even test solves add to the "players" total. Note that there are "3" players even though the puzzle is not yet published:

#12: Ki Bitzar (kibitzar) on Dec 27, 2018
Unfortunately the counter going up is true also for the user's solved puzzles. If one finishes a puzzle several times, the counter goes up. I have "solved" about 30 times your temporary test puzzle, Joe.
#13: Aurelian Ginkgo (AurelianGinkgo) on Dec 27, 2018
Good investigating, Detective Joe.
#14: Joe (infrapinklizzard) on Dec 27, 2018
I'm not sure what you mean, Ki.

Both the puzzle's "completed" count and the user's "puzzles solved" count get incremented by one ONLY WHEN the completed flag is *actively being set* to true in the database.

Note that the completed flag is not necessarily set each time you solve the same puzzle. If you've solved a puzzle and then you solve it again, the completed flag is not set(verb) because it is already set(adjective).
To give a simile: it's like counting flags being handed out by a referee to winners. If someone wins, the referee comes over and puts a flag in the winner's flag-stand. However, if they win a second time, the referee comes over, sees the flag-stand is full and does not hand out another flag.
#15: Aurelian Ginkgo (AurelianGinkgo) on Dec 27, 2018
Now I have a question: Do the stats reset every time a new version of the same puzzle is put out?
#16: Koreen (mom24plus) on Dec 27, 2018
So FYI, I like to keep up on what is newly being done on here, but since I am relatively new, there are a chunk of past puzzles I have not yet had the pleasure of solving. So sometimes, if I'm interested in a puzzle, and don't want to forget about it, I will click on it and save, then come back to it later. since I am sometimes off for weeks at a time, I may save a whole bunch. And I may not go back to those saved puzzles for quite awhile. Especially if when I'm back during some really interesting CURRENT puzzles to solve!
#17: Norma Dee (norm0908) on Dec 27, 2018
Considering the very large number of people on this site, a large chunk of the difference could be the people that are in the process of solving.
#18: Joe (infrapinklizzard) on Dec 27, 2018
Aurelian: I tested that, too when I added a version to 31914 and found out that it does not reset the stats.

It's probably important to remember that Jan put these numbers together as an interest point, not as a fool-proof, unimpeachable source of hard data.
#19: Aurelian Ginkgo (AurelianGinkgo) on Dec 27, 2018
That's what I'm thinking, Joe. They make a decent approximation, but not a good precise statistic. At first, I was thinking after your verdict, "What's the point then, if it's inaccurate?" But then I thought on it, and it is nice to know even an estimate if that's all we've got to go on. I think, therefore, the better number to focus on is only the completed stat, rather than the number started.

But at the end of the day, to me, it's just kind of cool to know. I don't think this discovery is going to bring about any sort of earth-shattering epiphany that affects my life decisions. So however the stat works is fine with me.
#20: Ki Bitzar (kibitzar) on Dec 28, 2018
I thought the discussion is covering only the stats that are visible from the creators pages. And wanted to point out that the same can happen on the users side.
#21: Brian Bellis (mootpoint) on Dec 28, 2018
I'm glad to have started this conversation. It has been very interesting. I think that Aurelian has a good point. These stats are interesting but should not be taken too seriously. Joe, I like how you performed experiments to check out all these hypotheses. Well done.
#22: Joe (infrapinklizzard) on Dec 28, 2018
Ki: that's most of the process. I did not want to lay out HOW to do it (let people who want to screw up things figure it out for themselves). As you can see from comment #7 I already pointed out that it can be done and gave proof. What confused me was that since I had already pointed that out, it sounded like you were trying to argue with my conclusion.

Brian: I agree with your agreeing with Aurelian's agreement with my opinion of the stats.

On a side note, There are over 8500 registered users since 2012 who have solved exactly 0 puzzles. There were probably several thousand before that, but Jan last purged the database in 2012. So each of them probably loaded several puzzles before giving up. Plus the innumerable unregistered people who stop by, look at a couple and then mutter "these people need less time on their hands" before heading off to facebook.
#23: Wombat (wombatilim) on Jan 4, 2019
Might be worth noting that you can solve puzzles without creating an account, but if there's no account there's nowhere to save the "completed" marker to... so to circle back to Joe's research, it would also be a way to increase the "play" count without increasing the "completed" count.
#24: Joe (infrapinklizzard) on Jan 4, 2019
You're absolutely right. I just created Yet Another Test Puzzle. Then I opened it multiple times in incognito mode. Yup. Even for non-logged in users, one page load = one new "player". And solving it definitely does not add to the "completed" tally.

It's been so long I didn't even remember that solving a puzzle while not logged in doesn't even come up with a "Yahoo!"
#25: Teresa K (fasstar) on Jan 5, 2019
No Yahoo for anonymous players? How sad. I remember the first time I saw that and was so thrilled. (✬‿✬)
#26: Philip (Philip) on Jan 5, 2019
Yes, I played for a while without creating an account and didn't realize how much I was missing.
#27: JoDeen Mozena (ozymoe) on Jan 5, 2019
I cannot imagine a person who is like-minded about puzzles such as these...and who is not absolutely thrilled to discover all you diverse AND similar minds/personalities who add so much interest and fun (and cleverness) to an already enjoyable solving process.

Thanks for being part of my consciousness...thanks for being part of my life.

I appreciate the way in which you are trying to figure out the stats, Brian, Joe and Aurelian. Me...I solve everything I can by logic. If I don't finish a puzzle it is because I intend to return to it with renewed vigor someday...which I do regularly.

Goto next topic

You must register and log in to be able to participate in this discussion.