Fastest solvers: An experiment

[Eureka]

That's quite an ambitious project you've got going there Klebbtheklari. It will be interesting to see the results. I'm guessing that you, will and azhang are all Science Olympiad friends?

[puffybob]

Tremendous, Klebbtheklari. Can't wait to see the results.

[Klebbtheklari]

Inspired by all of you (and being online friends with will and azhang), I have embarked on an even bigger (crazier? stupider?) experiment related to this.

The Giant Crypto Experiment, Continued to Top 3's

I am collecting the names of the top 3 UNIQUE solve times from ~500 Cryptograms. (yes there are lot of names...)
I will then condense the data using something along the lines of: each 1st place finish will yield 3 points, followed by 2 points for 2nd, and 1 for 3rd.
Any and all repeats will be included.
Any top 3's that I achieve during this period will NOT be included, though they will be noted.
This will (hopefully) be done at 100% solve rate.

Hopefully this goes well.

As of writing this post, I am 100 solves in, and there are already many clear patterns.

Data (which is being updated live!):
https://docs.google.com/spreadsheets...it?usp=sharing

Why do this?
I noticed a lot of recurring names in the higher spots of leaderboards, but these people rarely got records. So, I wondered if there were any implications from this and hence extended this experiment to more spots.

[nedzeppelin1453]

Wow, I never realized just how many records the top few players have. Thanks to everyone who collected data. It would be great if admin could keep track of records, in monthly competitions or even better, as another all-time ranking list.

[will0416]

Ok, thanks Eureka! I'll make the edit.

[Eureka]

Will/mysavinggrace, in case you missed it in the chat box, munchlet and MoiAussi are the same person. Many thanks to you and azhang941 for compiling these lists. It makes my job as archivist that much easier.

[puffybob]

Great stuff. I'm always just pleased to see that I still have even one record out there! I would love it if the site could somehow tell us how many records each player holds.

[Deanna48]

Thanks for doing this guys...it's neat to see the results! Of course, if any of us want to retain any records at all, we will have to get Will back to school!!! Maybe summer camp???

[gavinl]

i'm happy just to crack the top 20 list. It would be interesting to know though, the list of record holders. I have often wondered about that myself. Maybe a new section on the "High Scores" page could be created for everyone to see?

[azhang941]

I agreed earlier to do 500 more on top of will's 500 so we can have 1000 results similarly to puffybob's original experiment. I did keep track of ties, and I didn't get the result for one cryptogram because I double hit enter, but otherwise this was done at a 100% rate. I also did not include any records I set in this count.
Also quick edit: I included repeats. There was at least one game that I was pretty sure it was a repeat, but I left it in anyways because I couldn't tell with 100% certainty if it was a repeat or not.
will0416 105 + mysavinggrace 30 = 135
nanrich 107
bansaisequoia 86 + bonzaisekuoia 1 =87
lacrossesticks 67 + lacrosseballs 2 + lacatorcesticks 1 = 70
munchlet 49 + MoiAussi 5 = 54
twskinner 42
Deanna48 33
Descifrador 13 + JCarlosCuevas 5 = 18
Grambow 12
Altoid701 12
defeaterdefeater 11
azhang941 9 + SyoCB 2 = 11
DMVJDCPA 7
rotorguru 5
Shaquille 3
Lurker 3
Zachman285 2
Annamariah 2
xenia 2
bbray 2
caa 1
puffybob 1
laura 1
kat 1
nedzeppelin1453 1
pajarito7 1
kb83 1
Disturbed1 1
wings35 1
gryhnd51 1
damsel22 1
FloatLikeButterfly 1
peaceandluv87 1
Unwinding_at_The_Falcon 1
pootie49 1
LilyH 1
Cheese555 1

A full spreadsheet of the exact results I got in order and also a page with not including ties can be found in here: https://docs.google.com/spreadsheets...f6w/edit#gid=0

[mysavinggrace]

I was curious how the atmosphere of the site had changed in the past couple of months, so I conducted a sort of mini-study myself too (pretty much a continuation of puffybob's thing). Here are the record tallies that I found over the course of 500 puzzles (disclaimer: I skipped ~4 puzzles because I couldn't bear sitting there for over a couple of minutes, but I don't think the bias introduced by those will be too noticeable). I made two lists: one showing the original record holders of the puzzle (a.k.a. not including my most recent attempt on them) and the other showing the record holder after I was done with it (the former is definitely a more accurate depiction of the spread of records on the site). I didn't count ties, just the username that popped up as the record holder (for simplicity's sake) and also tried to keep any pseudonyms in mind, grouping the amount of records by person rather than by username. Please let me know of any pseudonyms I left separated.

List 1

will0416 - 120
nanrich - 92
bansaisequoia - 75
lacrossesticks - 52
twskinner - 36
munchlet - 29
Deanna48 - 18
Descifrador - 12
azhang941 - 10
Annamariah - 8
DMVJDCPA - 7
GramBow - 5
damsel22 - 5
Altoid701 - 4
Lurker - 3
defeaterdefeater - 3
kb83 - 3
contrary - 2
patsy86 - 2
rotorguru - 2
wings35 - 1
xenia - 1
mellav1 - 1
debzhaus - 1
puffybob - 1
jimdgar - 1
DivideByZero - 1
steakkguy - 1
nedzeppelin1453 - 1
Alduto - 1
Cheese555 - 1
jliu10000 - 1

List 2

will0416 - 313
nanrich - 51
bansaisequoia - 35
lacrossesticks - 29
twskinner - 17
munchlet - 10
Deanna48 - 8
azhang941 - 5
DMVJDCPA - 5
Descifrador - 5
Annamariah - 4
damsel22 - 3
GramBow - 2
Altoid741 - 2
kb83 - 2
wings35 - 1
puffybob - 1
jimdgar - 1
defeaterdefeater - 1
nedzeppelin1453 - 1
Cheese555 - 1
jliu10000 - 1
rotorguru - 1
DivideByZero - 1

The only record that was slightly bittersweet to break was steakkguy's. He was a great mentor for me and his 48 second record stood the test of time for 2 years on one of the more unusual quotes I had seen here. Hopefully I see his name on this site again.

Oh, and for anyone who doesn't know/is curious, I'm will0416 (this is my 2nd account).

[Eureka]

Thanks for sharing this, puffybob! What a lot of work for you, but very interesting to see. I agree with you, LLapp, Deanna and oddcouple regarding the fast times. I've solved a few puzzles in 8 seconds, and I am a slow-paced, middle-aged person with a poor memory. However, I do see patterns that help me recognize words right away and I have good typing skills and a fairly fast internet connection, which helps a lot. Some of our current young players are in their early teens, and have extremely good typing skills and sharp minds. If I can finish in 8 seconds, I'm not at all surprised (but am very impressed!) that they can finish puzzles in 3 or 4 seconds.

[LLapp]

Hello Susan Root -

Artificial intelligence would solve a cryptogram in 1 or, at most, 2 seconds, and not in 7, 8, or 9 seconds, which is what you see from the fast players here. I've seen discussion from longtime players who recall that there was a player long ago who did use a solving program, and that player's solving times were actually 1 second; that person and all their records were taken off the site.

Be assured that it is well within human ability to solve a cryptogram in under 10 seconds on this site, especially since the auto-fill feature means you type each included letter of the alphabet only once. Given auto-fill, if a quote uses, say, 20 letters of the alphabet, and you type those 20 letters in 10 seconds to solve it, then your typing speed would come out to 120 letters per minute -- which equals only 24 words per minute (assuming the average word has 5 letters). So even a slow touch typist could solve a cryptogram here in 10 seconds, especially if the letter patterns are obvious and easily decoded. The easier the puzzle, the more people solve it in under 10 seconds. I've seen some puzzles here where the chart of "top 20 solvers" shows speeds no longer than 8 seconds.

You are not the first player to express certainty that some players must be using an AI program to solve cryptograms so fast. If you stick with it and keep practicing, you will see how the fast speeds are quite doable. I started 5 years ago with an average solving speed above 70 seconds; I've since gotten my average speed down to about 31 seconds, and I've broken 10 seconds many, many times.

- Lisa

[puffybob]

I agree with Deanna48 and oddcouple. At first, I too found the record times hard to believe. But after a while, you get better and start coming close to them yourself. Even with my lousy typing skills, I'm often only three or four seconds behind the record holder. This was one of the benefits that came out of listing the top 20 finishers on each puzzle: you could see that a lot of people are getting pretty fast times. I suppose there's no way to prove that there's no cheating going on, but if someone really were using AI, you'd expect them to be excelling on every puzzle they do, when in fact the "top 20" listings show that everyone *sometimes* has scores that are significantly behind the record-holder.

[oddcouple]

Yeah, I've had about 40 or so sub-10 second times and I am definitely NOT a bot. BTW, I've NEVER set a record with any of my super-low times.