croco-puzzle

[sf2l]

a question for Berni or Calavera
which is the difference between "initialisieren" and "starten".

I thought the time started only on pressing on START. however today I clicked on INITIALISIEREN, then went to the alte Priesraetsel to practice a bit, then went back, pushed START, and the time was taken from the very beginning. A bug or a misunderstanding?
stefano

[ronaldx]

'Initialiseren' starts the timer for the day's puzzle (on the server).
Even if you quit the app, go away, and come back, your time has already started counting.

[Calavera]

'Initialiseren' starts the timer for the day's puzzle (on the server).
Even if you quit the app, go away, and come back, your time has already started counting.

This is true, but as far as I understood it, the server time is a backup which is only used when the user is closing the applet after clicking "Initialisieren".

If this is not true and the time is always taken from when "Initialisieren" is clicked I definitely have to change my starting pattern as I always try to calm down myself after clicking "Initialisieren" but before clicking "Start" .

[euklid]

It is as Calavera mentioned: After pressing "Initialisieren" the server clock is starting to run. Furthermore the puzzle information is loaded onto the local computer such that different internet connection speed does not matter. After pressing "Start" a second clock is starting in the local Java applet. This second clock is used for the timing usually. But if you close the applet after "Initialisieren" then the backup time (server timing) will be taken.

Thus: Don't close the applet after pressing "Initalisieren".

Have fun,
Stefan

[dickoon]

Congratulations to Nick for being awarded 20th kyu!

I don't propose to do this for everyone on every grade as that would get boring, but 20th kyu is a big one.

[nickdeller]

Thank kyu!

I've been pondering, ponderously... if we take into consideration the group of players DavidMcN, detuned, dickoon, oenomel, PuzzleScot, rodders, ronaldx, when do we think we'll first be celebrating two new kyu/Dan awards on the same day? I think it'll happen before the end of February, though I wouldn't like to speculate which pair will be involved.

[PuzzleScot]

Yup, well done Nick! I'd noticed you were close last night, but didn't think you'd sprint over the line today!

Hmm - a rating system within a rating system - obscurerer than ever!

[nickdeller]

It's fairly easy to guesstimate at this stage - we'll all be dropping zeros from the back end of our 200-Durchschnitt for the first seven months or so (assuming a few missed days in that time), so for that initial period it goes up each day by (Current Rating/200). On that basis, I'd expect the 20.Kyu to 19.Kyu transition to take 20 to 25 days for you, me and Ronald, with Ronald perhaps being the fastest by a day or so. I suspect Tom and David will already be pushing a 1000 rating by the time their 200-Durchschnitt catches up with them, so they may well need less than 20 days to get from 20.Kyu to 19.Kyu.

[nickdeller]

I'm too impatient to wait overnight for it to go formal, so congratulations to ronaldx on joining the ranks of the kyued!

[ronaldx]

Oh! Thanks Nick !!

[PuzzleScot]

Ditto. Was going to save the praise for a dual congratulations on breaking into a 4-digit score, but that might have to wait until xmas eve... Well done!
I see there are some people flying up the board, and some of us just staggering up.

[PuzzleScot]

Today's puzzles are repeats. I did the pills faster than last time, but included (000) as an extra pill, incurring a you-know what...

[nickdeller]

I got the most annoying you-know-what ever on the Pillen - multiple-clicked the centre of the final pill back to blank by accident, and had already hit Testen in a reasonably competitive time before the error had reached my screen. Oh well - the Masyu seems to more than make up for it.

[PuzzleScot]

Pah. They updated one of them. Restarted twice, then still made a dumb ..urgrrr.. on the masyu. On paper, that's a 60 second puzzle!
Oh, DM just flew past me - everyone else watch out! Tom, you're doing great!

[detuned]

Haha - the amount of pain I've been through writing some of my masyu puzzles has made most of the tricks in a small sized puzzle second nature by now! Unfortunately I was interrupted with a phone call during the pillen, but I'm making a point of having a 100% completion record, even if it means a few negative points here and there!

I must say, I'm enjoying the fragen zum rating thing - presumably a short term indicator - that's put me at 1900 now!

[ronaldx]

FzR has me at 1799 now (which is the highest I've seen it)
but Tagesrating has me working for the last 30 puzzles at 1k-2k level = 1950

Which is more realistic?

[Calavera]

FzR has me at 1799 now (which is the highest I've seen it)
but Tagesrating has me working for the last 30 puzzles at 1k-2k level = 1950

Which is more realistic?

Probably the latter one. My estimated rating was 200+ too low. At that time the Tagesrating didn't exist so I can make no guesses about it but FzR doesn't take into account that you get better once you are used to the puzzles and interfaces.

[PuzzleScot]

Great going one day, then very reasonable times ruined by silly mistakes the next day. (Not actually sure how that extra '2' appeared in the 2nd puzzle between me finishing it and clicking 'testen' - pretty sure I didn't put it there. Oh well)

Sadly croco-puzzle seems to have become a 'last thing at night' exercise rather than a 'find time during the day as a diversion' thing. Too addictive!

Oh, all the best for 2011 to everyone...

[ronaldx]

What's the deal with the account named 'ufo'? Has it affected everyone's scores?

[Calavera]

What's the deal with the account named 'ufo'? Has it affected everyone's scores?

When the top solution time changes it always affects everyones score. My guess would be that it's a second nick of someone (uvo already stated he has nothing to do with it). As berni is currently having holidays it might take a few days till he get sorted out. But if he cheated he'll be deleted as soon as berni is back.

[ronaldx]

Could be wrong, but it has the feeling of a duplicate account (increasingly). Hope Berni is enjoying the holidays

[PuzzleScot]

What's with Doppelblock? sum 1-8 = 28, but some sums >28! Urgh - should do these in the daytime. (No edit - just confessing my stupidity)

[ronaldx]

Oh well, at least you've obviously got a good idea of the puzzle

[ronaldx]

I notice that "anqu" has also come under suspicion on the German forums, for usually coming last but occasionally coming first.

That's just weird.

[motris]

I have no doubt that both anqu and ufo are cheaters, most likely as second accounts. For the case against anqu, look at the following:

12/18/10 - anqu scored 50s on a puzzle that took the 2nd solver 3m38s. At the same time anqu was last (42 min, with 7 errors) on the other puzzle.
1/3/11 - anqu scored 52s on a Doppelstern that took 2nd place 1m27s, while getting ~300th rank on the other puzzle the same day.
1/5/11 - Today uvo beat the cheater, but again very high rank on one puzzle and almost last rank on the other.

These results are not possible by probability to be such large outliers, in both directions, on the same day consistently. This is most likely a punk who has 2 accounts, or has hacked the software in a way that prevents accurate logging of time data.

Similarly, ufo has only played 3 puzzles, but has been first on all of them, and twice (both 1/4/11) by times well below those of people who aren't just copying an answer. 26s on a puzzle that takes 2nd place 1m47s is impossibly fast save for cheating - "typing a solution". There is no "lucky" in a Magische Labyrinth like there could be in Doppelstern, for example. This is again blatant cheating. I'd ban his account right now.

Cheaters will always exist in competitions, even when there are only nominal prizes, and identifying them on a case by case basis is simple when they do it blatantly. But "subtle" cheaters - not Eugenes - are always my concern. This makes me wonder if small but to me sensible changes in the ranking scheme are good for all solvers but also to remove effects of cheaters. One simple approach is to not allow the top rank to come from someone who isn't a ranked solver, say at 10k or better. Sure, this disqualifies me for awhile, but it prevents a first-week user, unlikely to actually solve that well, from affecting everyone's scores. A more involved approach would be to fix the ranking to be a more robust system. Using a single score (which could be anywhere from a 2 to 10+ s.d. outlier depending on how low the uvo time is, or if someone is cheating) seems less than ideal. The presence or absence of a single solver should not affect the whole range of scores, when having 300+ time results gives a good sense of the distribution. If one doesn't want to get into means and s.d.'s (which is what I'd do), then let's propose this. The 99th percentile is a 3000 rank. This is effectively 3rd place. Everyone on the podium gets a 3000 and times normalize down from there. The 3rd place score on a puzzle will be a more stable ranking of a puzzle's front end "fastest" time, than the 1st place score.