Colorful lawn or garden fixture

Elizabeth A. Long had a couple of Monday puzzles published last July and today’s Thursday, July 17 crossword (answers) is her first late-week entry. It has the kind of gimmick that makes you question all your answers until you figure it out.

Whenever all the theme answers are in Down clues, you know something is, so to speak, up. It’s harder to read Down so all things being equal, the interesting answers are always Across. The last Downy example I recall was back on July 9 when Tim Wescott needed the flexibility to put his math answers in the right boxes. Today’s downers are required by the theme which invites us to “quit misbehaving” or SHAPE UP.

This has consequences for the fill too. If you create a puzzle where a percussionist plays the ELGNAIRT, you have to give the solver enough help to push through, especially if it’s only Thursday. This puzzle manages an excellent balance.

BELLE Starr of the Old West was, by all reports, a very tough outlaw. Her murder is still unsolved, if you’d like to tackle a different kind of puzzle. I’ve got Sondheim’s song “The Worst PIES in London” running through my head now. More likely to cause slipping is ICIER although it’s not the ice you slip on but the water that forms when the ice is under pressure. If it's cold enough, icier gives you more traction. James AGEE is a common crossword name. He wrote The African Queen screenplay. Experimental filmmaker STAN Brakhage is obscure but the crosses were easy enough so I got to learn something. I’d spend more time watching experimental films if I weren’t such an ERAUQS. Clifford Odets gets remembered for something other than his first play Waiting for Lefty. Queen ANNE, the last of the Stuarts, succeeded William and Mary in 1702 and, five years later, became the first monarch of the united Kingdom of Great Britain. My favorite clue and the only one to make me laugh out loud is another short one — "Mobile home." Click the answers link if you haven't solved it yet.

The theme answer that nearly confounded me was RATS. Since that’s a legitimate word, I spent way too much time wondering what rats had to do with headliners before finally remembering the theme. Rats, indeed.

I've had some questions about the XWord Info Freshness Factor.™ The formula is explained on the site (click the Freshness link from any puzzle) but the basic idea is that if every answer is unique, the puzzle scores 100. Rare words drop the score a little and very common words drag the score down a lot. Today's puzzle has a rather low score despite 6 unique words, reflecting the need to keep the puzzle simple enough to withstand the thematic oddness. Any puzzle with ERA, ERIE and ALA is going to score poorly. Scoring 70 is fresh, 80 is piping hot, and only 12 puzzles score over 90. The highest score in my database is an amazing 96.4. High score doesn't necessarily mean great puzzle, of course. One of my favorite puzzles scores the lowest: 28.0.

An interesting feature of the Freshness page is that you can see at a glance all the answer words unique to each puzzle. I might not have guessed, for example, that yesterday's MERRIER and ROADHOG were both debuts. More on Freshness in an upcoming post.

Comments

[KarmaSartre]

That was a kick. I used the giveaway answer to help figure the theme. With SUED at 65a, and what little I had gotten of the upward clues.I figured 42d ended in UP, and that was all it took. Very enjoyable theme to figure out. Wasn't sure about the LORELAI spelling (with the A), and didn't know DINAH, so had to guess. I never heard of a PC running on MSDOS, though MSDOS ran on many a PC...that felt backward. I agree on "Mobile home": nicely clever.

[PhillySolver]

Jumped around and got to the theme clue and saw I didn't have any of those filled in. Went to the Race Track and just needed one letter for OVAL (LAVO to you). Still took me awhile because EPISTLE and ARTICHOKE wouldn't come to me and I ended up felling out of EPAHS. Which reminds me of the Musical Grease and the fun song...'You Better Shape Up.'

[Orange]

Jim, will you have a page that lists, say, the Freshness Hall of Fame? That'd be cool to see. And can you calculate a constructor's average freshness?

Granted, rarity of a crossword answer is no guarantee of quality. An obscure kind of turpentine, for example (TMS crossword!), is not welcome freshness—but Will Shortz isn't in the habit of letting the bad kind of fresh in.

[Jim Horne]

The Freshness Hall of Fame.™ I like that.

I've got such a list ready to go but I'll give you a few days to think about it. The first details will be published Sunday night as part of my Monday puzzle post.

[joon]

i like the freshness feature (hooray for random comments on super-old threads), but i'd like to suggest a modification in the way it's calculated. as it stands, freshness depends way more on the stalest entries than it does on the freshest ones. for example, a unique entry is only worth a couple more points than an entry which has been around a few times, but you get heavily penalized for using, say, ERA (freshness: -220) as opposed to ARE (freshness: -124). do those two really feel qualitatively different to you when you think about how fresh a puzzle is?

so i'd suggest either a cap on the staleness (maybe freshness can't go below 0?), or maybe some sort of smarter average that weights the fresher entries much more heavily than the less fresh entries. one formula (off the top of my head) is to take the # of occurrences for each word (including that puzzle itself, so the minimum is 1) and then do a geometric mean. then you could subtract from 100.

as for why the geometric mean instead of a normal average (arithmetic mean), here's a simplified example: say you have a four-word puzzle. if three entries are unique and the other has been used 81 times, that feels quite a lot fresher to me than a puzzle with two unique entries and two that have been used 41 times each. they have the same arithmetic mean, but the geometric mean of the first puzzle is 3 (fresher) and the second is sqrt(41), or about 6.4 (less fresh). that seems right to me.

sorry to go all number-geek on you. i spend a ton of time thinking about stuff like this (metrics for various kinds of performances, often baseball-related).

[KarmaSartre]

I'm enjoying this numbers talk. I'm thinking bill james.

[ArtLvr]

Great blog, great discussion -- and Jim, you're right about how much more "visible" the across answers are, compared with the downs. Example: I was nearly done, knew I wanted SQUARE going upward, yet when I typed it I somehow let the LORELEI (no A) dominate!

I think joon has a good point, especially if you are rating Freshness. And will you have to give extra points or new category for the words that are deliberately misspelt or upsidedown or backward, or missing a letter (e.g. the non-U answers "Popl'ar Mechanics" or "Stat'e of Liberty")

[joon]

KS, it's a very jamesian idea. i believe he uses the geometric mean for calculating his "power-speed number," for much the same reason: a guy who's 20-20 is more of what we think of as a "power-speed" guy than somebody with 70 HR and 1 SB.

[lou45419]

Enjoying the new freshness index and accompanying discussion. I submitted my first xword recently and of course was curious how it would stack up. Not too bad just looking at the numbers, also uses A thru Z, and the scrabble score is not too shabby. It is interesting to hear what others look for in a puzzle or how they think "freshness" should be considered. I appreciate the different ways you give us to try to objectively assess a puzzle.

I had trouble getting the theme from the SHAPEUP clue. Focusing on the "shape" more than the "up" and needed to be paying attention to them both! Last puzzle liked this that I worked was in Amy's book and it had two gimmicks going. If you want a challenge and haven't done the Aug 03, 2006 NYT xword by Nancy Salomon give a shot.

Thanks for the Freshness factor along with everything else here Jim!

[lou45419]

Forgot to mention, Here is a quick way to figure a puzzle's Scrabble score.

[john farmer]

Interesting discussion. I agree with joon that the Freshness score is not perfect, but I'm not sure it needs the Jamesian treatment. I had been thinking of ways that the scoring could be refined (giving different weighting to long vs. short words, for example), but I think the stat might be just as useful as it is.

Before sabermetrics there was statistics. A baseball hitter gets an RBI for driving in a run. A solo home run and a sac fly are both worth one RBI, and they're hardly the same accomplishment, so the RBI stat is not a perfect measure of performance either. But it's useful. It's easy to understand, and that's a virtue.

In today's puzzle, TWEETS and WHIRLIGIG each score the same (1 other occurrence). Same for PRAISES and ARTICHOKE (3). In yesterday's, MERRIER (0) scores better than THUMBNAIL SKETCH (2). But in each case, I think the latter is better fill.

That's all fine, if you keep in mind that the score is only a measure of freshness, i.e., the frequency that a puzzle's answers appear in other NYT puzzles. It's not a measure of how lively the fill is.

No matter how much you refine the Freshness score you're never going to get a measure of the quality of the fill. If you want to measure the quality of fill, you could assign a score to each word and do some calculations. There are programs that can help you do that. But in the end, scoring is subjective. Quality of fill is more like the cleverness of a clue or the novelty of a theme: you may know it when you see it, but it's hard to quantify.

The Freshness score is a handy way to quantify one aspect of a puzzle. It's not everything, but it's useful, and I think it may help to keep it simple.

[alanrichard]

Of all the guys who post blogs and very good accompanying videos I thought you would have a Video of Rocky And Bullwinkle and the upsidasium.

[joon]

john,

you have a good point, but i wasn't talking about liveliness or what constitutes "better fill" or any other subjective measure. i was only talking about the technique used to aggregate freshness scores for individual words (which are easy and unambiguous to calculate) into a freshness score for an entire puzzle. the only difference between my idea and jim's is that he uses an arithmetic mean, and i use a geometric mean. i think the geometric mean makes more sense, and it's not really more complicated--it's just a different kind of average, which emphasizes the fresher entries instead of the staler ones.

jim's method is fine, obviously. but i think that it's more useful for identifying the stalest puzzles than the freshest ones. if you're going to start talking about a hall of fame, though, i demand a more advanced metric. :)

[PhillySolver]

I give Jim a freshness score of 99, joon a score of 98.9 and combining the two gives me a 100. That would be an 'accelerant' mean. There doesn't seem to be any place else that this discussion is taking place, so kudos to all.

I tried creating a "Difficulty" score and abandoned the idea after various permutations I just couldn't figure out a way to automate the calculations. I scored points for .var and ? clues and employed an originality scale as well as word plays identified by rhetorical device. I did prove that Saturday was generally the most difficult, but that wasn't earth shattering. The freshness score discussed here is for conversation and joins the scrabble score and pattern analysis as a detail that puzzle lovers appreciate. Thanks for all of those things, too.

[john farmer]

joon,

As I was thinking more about this freshness thing (waiting in line at FedEx today, what else was I to do?), I had second thoughts about some comments I'd made. I didn't mean to discourage any attempt to improve the way scores are calculated. You had some good ideas, worth considering. It would be interesting to see if a more refined measurement changes the scoring much. Perhaps in a few cases. In any case, whatever Jim decides to do, I'll be interested to see the results.

[Chad]

I am so glad I found this site! haha I had tried on a whim to solve this puzzle but was so confused I didn't know what to do! I mean I knew I had LORELAI, my mom works in a grocery store and gave me CASABA, I had CURE, RUNG, CRINKLE, ANNAN...IT WAS SO CONFUSING! haha thanks for enlightening me to the "theme" idea behind crosswords. Never heard of that before.