Wordle?

[xilman]

Originally Posted by Uncwilly

Ideally you guess 3 words with unique letters. That would get you info on 15 letters out of 26. That would narrow the field of possible words quite drastically.

That was my approach. I am far from certain that it is optimal but it is surely not pessimal.

Precisely which three words are best is still an open question AFAIK

[firejuggler]

First word is always Adieu

[jwaltos]

Originally Posted by xilman

If you don't know who Alec is, a search on his name and "crack" might prove informative.

I did not. I did. It was.
In the late 70's early 80's I used Solaris while studying eng./math in Windsor and worked with Sun and Apple machines in the early 90's, (eng./math again) at UBC. At the time, some smart cookies there successfully exploited a particular weakness in a BC lottery (quicky removed from the market). The math librarian there was a font of interesting information as well as a phenomenal archive spelunker (read Blissard's papers there after thinking I had come up with something new) UWaterloo is where I first came upon Knuth's Stanford Graphbase (and other writings) which were an eye opener. Good puzzles derived and contrived are always worth a look..Dudeney, Lloyd, Graham, Carroll..
https://terrytao.wordpress.com/2012/...logic-puzzles/

[chalsall]

Originally Posted by jwaltos

In the late 70's early 80's I used Solaris while studying eng./math in Windsor and worked with Sun and Apple machines in the early 90's, (eng./math again) at UBC. At the time, some smart cookies there successfully exploited a particular weakness in a BC lottery (quicky removed from the market).

If I may please share...

For some bizarre reason, a key component in a "Best of Industry" Class 4/5 "Soft" Telephony Switch I was once involved with procuring and deploying about a decade ago or so had a 2U server running Solaris.

The provider kept telling us: "Now, it's ***very*** important that you firewall this, or else you'll be hacked!

During training in San Fran, I was given an unprivileged Unix account onto an equivalent. Which, "surprise", was back in the day when /etc/passwd still exposed data worth "working" to anyone who had access to the file system.

I copied the file to my laptop, which "cracked" it in less than a second. They hadn't changed the default password! It was the first guess in the dictionary!

Correlation does not mean causality... But...

While leaving the SF airport shortly after that training session, it was the first time I had ever had guns unclipped and held (but, importantly, not drawn) during any TSA "interaction". They did a *serious* review of everything I was carrying. Frankly, I am thankful more exploration wasn't done.

This is a true story (as best as I remember it).

Why was Solaris still in the equation? I have no idea, but I know to watch out very carefully for the "end-of-life" kit in all solutions proposed now.

[Dr Sardonicus]

Originally Posted by Uncwilly

Ideally you guess 3 words with unique letters. That would get you info on 15 letters out of 26. That would narrow the field of possible words quite drastically.

No. "Ideally" is your first guess actually being the word:

Originally Posted by Uncwilly

I got the current one on the first try.

[Uncwilly]

Originally Posted by Uncwilly

Ideally you guess 3 words with unique letters. That would get you info on 15 letters out of 26. That would narrow the field of possible words quite drastically.

decaf humor lints as the first three guesses would account for 85.9% of all letter usage in English, but not specifically for 5 letter words. There are 2345 word combinations that would work for that. So, there is further optimization, with regards to letter order.

[Xyzzy]

Originally Posted by Uncwilly

Ideally you guess 3 words with unique letters. That would get you info on 15 letters out of 26. That would narrow the field of possible words quite drastically.

It is much more fun to play with "hard mode" turned on.

[Xyzzy]

Originally Posted by firejuggler

First word is always Adieu

SOARE

[xilman]

Originally Posted by Xyzzy

It is much more fun to play with "hard mode" turned on.

I play that way without turning it on explicitly. Perhaps that is why I am doing so poorly (2 games, 2 wins, 4&5 probes).

[LaurV]

When I graduated the high school, my diploma project (we were studying at a computer science high school, which was "something" at the time, only "clever" guys could get there , and had to present a project, thesis, or dissertation, don't know how to call it in English, for graduation) was about games theory, min-max, backtrack, blah-blah, and I had - as examples - Basic and Fortran programs for Mastermind and Go Moku. And yep, you had to load them from punched paper tape! The Go Moku program could not play very well, and it repeatedly went in endless loops after 15 moves or so (which was not discovered by the examiners, they were not really interested in such thing, and I solved much later during the early Uni years) but it could always win with black, and it beat most of the human players with white too. Futile to say that there were none of my colleagues neither in the HS nor the Uni that could beat me at that game! This was about 5 or 6 years before Victor Allis' proof (that the first player always wins, in free play) so I could also boast about the fact that I (somehow) proved it before him. Unfortunately, at that time, behind of the communist curtain, we had no idea that you could actually "publish" your work... (seriously!) and my work was not systematic like his. The fact that black always wins was not an unknown fact at the time, the Go Moku community knew it for centuries, that is why they invented Renju, which is just like Go Moku, but with some restrictions for the first player, however, there was no rigorous proof before Mr. Allis did his.

The Mastermind program was running on a M18 computer, whose output was a centronix console (yep, printed paper, all the dialog was printed, as the game progressed, line by line!) and in the beginning it was printing you the rules (like no duplicates, and zero can't be first) and asking "Do you want to play with four or with five digits?"***. Years later I presented the same program, this time in Pascal, on a tty* monitor, to one of my Uni professors, a very nice and clever guy, and I still remember the grin on his face. He intentionally pressed the button for "yes", and the displayed message was "Press 4 or 5, don't you know the numbers?". Then he lectured me 20 minutes, half joking, half very serious, about how to choose the messages that the program displays. The algorithm itself sucked, in the sense that all the arrangements were generated in lexicographic order and verified one by one, but this was well hidden with a lookup table (a random permutation of the vector 0 to 9 generated at the beginning of each game, and what was printed was not the arrangement itself, but what was in the random vector at those indexes). So, in theory, you could just chose no number for yourself, and reply a predefined sequences of answers** to the computer guesses (the playing was interactive, human and computer take turns at guessing each-other's number, like in chess, and not a single player guessing) and have the computer make the longest number of guesses, while you could be lucky and guess its number in less turns, therefore winning all the time. But in "correct" play, like you choose your numbers in the beginning, you will most of the time lose, because you would make mistakes (especially under time pressure and for the turns towards the end of the game, when you have more verification to do.

________
* at the time, that used to mean "teletype", and not "talk to you", like now :P

** such a predefined sequence would be that you reply "3-0" all the time, indicated that he guessed 3 digits and they are in the right position, so this would cause the computer to guess (assuming its random vector is the identity permutation): 1023, 1024, 1025, ... 1029, so he would need always 7 turns to guess, while you could guess his number much faster. Unfortunately (for you), this was considered in the program, and after a first "3-0" the computer would "guess" something like 1456, to split the remaining stuff, so you had to be more clever than that. This was very much appreciated by the examiners.

*** there made no sense to play with 3 digits or less, too simple, and also with 6 or more, which would transform the game in something like "just guess the position", etc., but 4 and 5 are nice to play, and 4 is easier than 5, so you could have some "easy" and "hard" levels to try

[LaurV]

Re Wordle, one can download TWL or SOWPODS, we still have them somewhere since we used to play competition scrabble. Filtering the five letter words is as simple as paste to excel, write "=len(a1)" in B1, click to duplicate downwards, click to data/sort by column B, that's all... (you may delete the other words or keep them, up to you)

We got the one today quite right from 4 trials (the fact that it started with A helped, hehe)

Edit: if you do that (i.e. sorting sowpods in excel), be careful of this:

Bwaaa... hahaha
(same for "true", you will need to apostrophe them both)
.