Speed Records and "Cheating"

[mockmook]

Howdy,

Before the migration there was a discussion about speed records and alleged cheating.

I'd like to throw in my two cents:

I suppose I am somewhat new here. I've completed about 300 puzzles, and kept track of my improvement with experience.

Anyway, my average time for solving a puzzle now is about 2.5 times more than the average time given on the puzzles.
But, sometimes I still struggle, have to use "Hints", and end up with times 5 times or more worse than the average.

And, it was just recently that I completed my FIRST (and ONLY to date) puzzle in less time than the given average.

Just for context: I am not a dolt :-)
I was beyond the 90th percentile when I used to take standardized tests.
I graduated college with honors for a BS degree -- yes, only a BS -- maybe I am a dolt?

So, how is it possible that my FASTEST time -- when I was filling in answers boom, boom, boom -- was still FOUR TIMES slower than the record for that puzzle?

I can see how I might sometimes get to twice as fast, but four times? Impossible.

But, that's not my main source of skepticism that everything is on the up and up. It is the AVERAGE time that mystifies me. The AVERAGE implies someone solving every puzzle in that boom boom boom manner. Little or no hesitation, knowing about 70% of the clues right off the top of your head -- then it is child's play to fill in the few missing squares.

Therefore, the AVERAGE solver routinely matches the best time I have ever achieved. I mean, are there no "novices" trying these games? What about the games where you leave a puzzle overnight and solve it the next morning? Doesn't that ever happen? (I know I have done it)

And, if the AVERAGE solver can get through virtually all his/her puzzles in a boom-boom-boom manner, how do they get any enjoyment from it? Part of my enjoyment is the struggle to find missing clues/squares -- I can't see how I would enjoy the games if every one virtually solves itself.

So, I am skeptical of the records AND the average times.

Though, I'm happy to hear other perspectives -- just know it will be a hard sell if you claim no shenanigans :-)

[compuspud]

I completely agree. I often solve the puzzle in about twice the record time. Even when I don't stop to think, I can't get down to 35 seconds as some often do. So, they either already have the answers and type them right in, or something else is going on. In fact, one of the top all time scorers has solved ONE MORE puzzle than he has played.

[11dotcom]

My wife and I play Wordtwist together most mornings, just for fun and to keep our brains exercised. We are quite proud of our best score of 574 points and our most words of 119, both of which took some doing. Yet when I look at the records, I see someone has scored 5552 points in one game and the most words is 805! How is that possible? It does make the idea of records rather pointless, doesn't it?

[bobmiller241321]

While there's a vast quantity of puzzles, there are occasions when you recognize a puzzle that you've completed already. That's where you get those super-fast scores from. Once you recognize the puzzle, you're only limited to how fast you can type.

[scitsorca]

Therefore, the AVERAGE solver routinely matches the best time I have ever achieved. I mean, are there no "novices" trying these games? What about the games where you leave a puzzle overnight and solve it the next morning? Doesn't that ever happen? (I know I have done it)

I only recently noticed the bellcurve at the bottom. I don't know how it is calculated. I don't think my puzzles ever go into the calculation because I'm not logged on when I do them. I just wanted to say that I routinely wander off to do other things in the middle of a puzzle. I mostly have a look in the morning and I get my coffee and let the dogs out and do that sort of thing or I go off up to the shops and leave the screen open. Other times, I think I'll look at a puzzle and have to go off to do other things so close out of it. Presumably, there are "diehard" puzzlers who log on and attempt a fast time? There's also a percentage at the top that says how many finished the puzzle - I think they'd have to be excluding unfinished puzzles from the calculation because (say) a 50% "failure rate" would drag down an average, wouldn't it?
I don't know, but I agree with you, I like to think about a puzzle that has me stumped until I "get it", I really don't care how long it takes me. I just ignore the time part. I suppose if you are really, really good at puzzles and have become familiar with the puzzle setters' style to get the answers very quickly, having a timed game gives you something additional to play for. As you say, answering every puzzle boom boom boom might get a bit "old" so seeing how quickly you can get the answers in might offer super-puzzlers something more than us "regular" types.
It would be interesting to know how they calculate the averages.

[compuspud]

I can understand a very low score on a puzzle you've done before. I can understand a very low score on puzzles where you happen to know all the answers right away. But to know all the answers immediately on all the puzzles is highly unlikely. Yet the same people always get the lowest scores. How does that happen.

[Dunderchief]

Compuspud, I know part of the answer. Some players are manipulating a game feature by repeatedly hitting the "Play" button until they get a nice short easy puzzle that they can get a very high score on. The player gets to select a different puzzle if he/she so chooses.

When you hit "Play" you get the puzzle's statistics, and you get to decide if you want to play that puzzle or not. You can tell if it is a longer puzzle or has a relatively lower percentage success rate by reviewing "This Puzzle's Statistics" , and if so you can simply hit "Play" again instead of "Start this puzzle", and if you do there is no penalty whatsoever to any of your personal statistics. And you can keep repeatedly hitting "Play" until you get a puzzle w/statistics to your liking.

Look at some of the High Score statistics for this month. The fastest player, jubl11, has an AVERAGE points of 799.3!!! The second, CiaraSchufrau, is at 796.6!!! Number 10 on the list is hootman, who is one of the current point leaders for High Score and who has played 847 games so far this month, and his average point score is 736!!! Halfdutch, at number 5, has played 623 games and has an average point score of 780.9!!! The absolute maximum on any puzzle is 800.

For anyone playing whatever puzzle comes up, long or short, hard or easy, these average scores are impossible. I have been on this sight since it began, and my scores are usually one side or the other of the line between "Very Fast" and "Fast". I know I am an excellent player. I have solved every single puzzle I started. I know I am not the fastest player here (JennyO, are you still out there? I still love you!) But my average over a total of about 8200 games or so is 567 points per game. So I know those averages in the 700's are impossible. I quit playing in competitive mode (mostly, anyway) a long time ago bcz the standings mean nothing.

RECOMMENDATION: It won't solve all problems, but if possible the repeat-play feature discussed above should be disabled for anyone playing in "Competition" mode. Anyone who hits "Play" should go straight to the puzzle. These average scores in the 700's would disappear, I'm sure. It is fine to leave this feature as-is for anyone NOT playing in "Competition" mode. A new player may well want to pick easier ones to get the hang of it. I generally like to pick longer harder puzzles as they are more of a challenge.

The other part of the answer is, as has already been suggested, some people play so many puzzles that they become extremely familiar w/often recurring clues and answers. The only thing I can think of to do about that is to put a limit on how many games any one player can play per month in "Competition" mode. I was thinking 150 games/month which is about 5 per day. This would not solve the problem entirely, but at least everyone could be competitive. The high score competition would not devolve into a who-can-play-the-most-games competition which most of us have no interest in. Who wants to play 50 games a day each and every day??? Takes the fun out of it. The only other possibility is to take hootmon and halfdutch and JpmUvm et all and put them in parachutes and drop them deep into sub-Saharan Africa, and may jackals feed well upon their bones!!!!!!

ADMIN: I hope you can respond.

[JpmUvm]

I'm one of those high-scoring players "Dunderchief" discussed. Although familiarity with repeated puzzle clues helps, it's not why my scores have increased and solving times decreased. It's almost entirely because one day I realized I could keep clicking the "play" button until I got a puzzle with an average solving time of under 400 seconds and a record time of under 120 seconds. I have received a score 6 or 7 times as high for a short puzzle in one-sixth or one-seventh the time I spent solving a very long puzzle. This means I can accumulate points 40-50 times as fast per hour! This is ludicrous. I actually miss the challenge of solving the long, harder puzzles more often than I do now.

The problem is the scoring system. It sets a maximum of 800 points, and then deducts points based solely on how much time it takes you to solve the puzzle, irrespective of its length. It reminds me of the old scoring systems used by gymnastics and figure skating at the Olympics. They would set a maximum score and deduct points for mistakes. This meant that no one could be rewarded for doing something new or unusually difficult. Now they deduct for mistakes from a minimum score and add points for "degree of difficulty." There might not be any maximum scores.

This website needs a similar scoring system. It seems to me a player's score should be based on 3 factors:

1) Average solving time of that puzzle compared to all puzzles. Could use a percentile system. A player should be rewarded for spending more time on a puzzle requiring more time.
2) Average solving percentage. Solving a puzzle that has a lower percentage should yield more points.
3) The percentage of an individual player's solving time compared to that puzzle's average time. The lower the percentage, the higher the score.

If a new scoring system along these lines would be adopted, I would become a paid member of this website.

[Andrew John]

Curious to see my scores now that I'm playing as a "member" after playing as a non-member for a year. I've beaten the posted high score once, but wasn't a member and couldn't log it. I don't know about how one might mechanically cheat, but some low scores have seemed humanly impossible. Once, I knew the quote from the first word that I put together (I knew the author) and typed the whole thing in as fast as I could and was still 30 seconds behind the lowest time. I've encountered quotes twice, recognized and remember them and typed them in, and still not got close to high score...

I'm usually at about one half to one quarter of the average time, and often within 30 to 60 seconds of the best time. No college degrees, just widely-read, a bit manic, and a writer myself. (I've encountered 5 quotes from the books of friends in the last year, and sent 3 of them screenshots; all were quite chuffed)

[Dunderchief]

JpmUvm, I like your thought to revise the scoring system.

You gave me the idea that maybe one or two points should be added for each letter in the puzzle, so longer puzzles would automatically be eligible for more points, in order to make up for the longer solving time. Or something along those lines. Something to give more points for longer puzzles. No more max-out at 800. I have no idea how easy or hard or practical any of these ideas about changing the scoring system may be.

And still, I would like to see the ability to select new puzzles removed for anyone in competition mode which you fessed up to doing -- you're an honest guy! But keep it active for anyone not logged-in to competitive mode. Let the newbies pick easy ones to learn on, and let guys like us pick more challenging ones.

I agree w/you that the longer harder puzzles are more fun. I (mostly) quit playing in competition mode but still play w/o signing in, and purposely select longer harder puzzles. I was into chasing statistics for awhile there in the old days, but grew tired of it once I realized how meaningless the stats are.

[Morticia]

I've found one way to get the super-fast times -- have something go wrong in submitting your answer. I've been disconnected several times, and it was just a few months ago that I discovered that, when something goes wrong, the game is saved for a couple of hours in "Recent Games", so you can go back to it and try it again. There have been a couple of times when the problem came as I submitted my finished puzzle. When I reopened the "Recent Games" one, it had me starting over, in letters and time -- on a puzzle I just solved. I could have gone through and answered all the clues, or filled in the quote, in 30-40 seconds -- just a matter of how fast I could type (and remember). I tried to spread it out over roughly how long it took me the first time, although human nature being what it is, I was probably faster. Anyway, this may account for some of the blistering-fast times. I like that you can get another chance if something outside your control went wrong, but maybe the resulting time shouldn't be factored into your total.

[redpicker]

...

This website needs a similar scoring system. It seems to me a player's score should be based on 3 factors:

1) Average solving time of that puzzle compared to all puzzles. Could use a percentile system. A player should be rewarded for spending more time on a puzzle requiring more time.
2) Average solving percentage. Solving a puzzle that has a lower percentage should yield more points.
3) The percentage of an individual player's solving time compared to that puzzle's average time. The lower the percentage, the higher the score.

If a new scoring system along these lines would be adopted, I would become a paid member of this website.

The problem with tying scoring to a puzzle's average times is that the average time changes each time a puzzle is completed. If a new puzzle has an infinite time (or, just arbitrarily assigned a very long average time, say 86,400 seconds or 24 hours), then its average solving time is greater than the overall average, so it's point value would be high. As the puzzle gets played, its average time would drop and its point value would drop. The opposite would occur if it is assigned a short average time (say, one second). The point is that you cannot have the scoring equitable if the point value of the puzzle is tied to its average time compared to the average of all since the average time of a specific puzzle will constantly be changing as more people play the puzzle, but the overall average not so much

You have the same problem with average solving percentage. If it's a hard puzzle, the average is going to go down as more people play it, so later players of the same puzzle will get more points than earlier players of the same puzzle. If it's an easy one, then earlier players will get more points than later ones, because the average keeps changing the more the puzzle gets played.

The only thing that makes sense to me is something like a puzzle is worth one point per letter in the puzzle, then if you complete that puzzle in less than one second per letter, you get full value. Each hint would detract one tenth of the point value, so if you used all 5 hints, the value would only be 50% of what it was. If you do it in two seconds per letter, you get half value, three seconds, one third value, ... , ten seconds per letter one tenth value. You could start out at 2 points per letter, or whatever. If a player clicks on the "Play" button without starting a puzzle (after seeing the stats on the puzzle), then that player is given a "score" for that puzzle based on a solving time of 30 seconds per letter (or whatever seems fair--it has to be enough to make it make more sense to finish a puzzle than just click through).

The other option is to just ignore the scoring. Play for the sake of playing and forget everything else. I could meet "Dunderchief" on the street and never know it (at least, not of his dirty deeds), so it's not like having a record-breaking score actually means anything.

[redpicker]

...

This website needs a similar scoring system. It seems to me a player's score should be based on 3 factors:

1) Average solving time of that puzzle compared to all puzzles. Could use a percentile system. A player should be rewarded for spending more time on a puzzle requiring more time.
2) Average solving percentage. Solving a puzzle that has a lower percentage should yield more points.
3) The percentage of an individual player's solving time compared to that puzzle's average time. The lower the percentage, the higher the score.

If a new scoring system along these lines would be adopted, I would become a paid member of this website.

The problem with tying scoring to a puzzle's average times is that the average time changes each time a puzzle is completed. If a new puzzle has an infinite time (or, just arbitrarily assigned a very long average time, say 86,400 seconds or 24 hours), then its average solving time is greater than the overall average, so it's point value would be high. As the puzzle gets played, its average time would drop and its point value would drop. The opposite would occur if it is assigned a short average time (say, one second). The point is that you cannot have the scoring equitable if the point value of the puzzle is tied to its average time compared to the average of all since the average time of a specific puzzle will constantly be changing as more people play the puzzle, but the overall average not so much

You have the same problem with average solving percentage. If it's a hard puzzle, the average is going to go down as more people play it, so later players of the same puzzle will get more points than earlier players of the same puzzle. If it's an easy one, then earlier players will get more points than later ones, because the average keeps changing the more the puzzle gets played.

The only thing that makes sense to me is something like a puzzle is worth one point per letter in the puzzle, then if you complete that puzzle in less than one second per letter, you get full value. Each hint would detract one tenth of the point value, so if you used all 5 hints, the value would only be 50% of what it was. If you do it in two seconds per letter, you get half value, three seconds, one third value, ... , ten seconds per letter one tenth value. You could start out at 2 points per letter, or whatever. If a player clicks on the "Play" button without starting a puzzle (after seeing the stats on the puzzle), then that player is given a "score" for that puzzle based on a solving time of 30 seconds per letter (or whatever seems fair--it has to be enough to make it make more sense to finish a puzzle than just click through).

The other option is to just ignore the scoring. Play for the sake of playing and forget everything else. I could meet "Dunderchief" on the street and never know it (at least, not of his dirty deeds), so it's not like having a record-breaking score actually means anything.

[Dunderchief]

I could meet "Dunderchief" on the street and never know it (at least, not of his dirty deeds), so it's not like having a record-breaking score actually means anything.

What dirty deeds? I protest - I am the one calling for the dismantling of the ability to use the keep hitting "Play" until you get an easy puzzle! I haven't played competitively since 2014 bcz the competition devolved into areas I was not interested in, the repeated-play issue plus the issue that you have to solve a huge number of puzzles to compete for high score. Except for a couple of months many years ago, I take all puzzles as they come up whether long or short, hard or easy. I take umbrage at "dirty deeds" no matter how lightly you used it.

[redpicker]

Sorry, no offense intended. Just a reference to that AC/DC song, you know, "Dirty Deeds and the Dunderchief". I just assumed that is where you got your handle.