xwordinfo analysis very high from computer generated puzzle

[bgreen]

I have searched through the forum and did not see this topic. Please let me know if this is redundant.

I was curious how the xword info Analyze puzzle feature would work, so I put random blocks in Crossfire, clicked quick filled, and uploaded it.

It scored very highly. Perfectly in line with all the metrics. My question is how do I know when I have an actually good puzzle? The puzzle I made in minutes scored as well as some of the best puzzles I have seen this year.

[mmcbs]

That analysis only looks at factors such as word length, letter distributions, numbers of entries and black squares. It also considers how many words you have that have not appeared in crosswords and the frequency of usage of your words. It makes no judgment about the quality of words in the context of whether they are interesting or meaningful to the solver. So it really doesn't tell you if your puzzle is "good" - it only tells you how it compares statistically to other NYT puzzles. It's not necessarily a good thing to have a lot of words that have never been used if they are not words that an editor would allow in a puzzle.

[mmcbs]

I think the main benefit of that analysis is to determine where your puzzle best fits in the spectrum of NYT puzzles (Monday-Saturday level for the 15x15 puzzles).

[bgreen]

Thank you both!