Visit Hot's Puzzle Page. Send me e-mail .

Cryptic Thoughts

bar diagram

by Hot (Henri Picciotto)

I wrote this manifesto in July 1992, mailed it to various cryptic constructors and editors, and handed it out at the San Diego National Puzzlers' League convention. Part of it was excerpted by Sibyl and included in the Guide to the Enigma. Throughout, I refer to NPL members by their noms. (For example, WILLz is Will Shortz, Hex are Emily Cox and Henry Rathvon, Manx is Mike Shenk.)

It is still not easy to have puzzles published if they deviate from the norm, but Henry Hook's and Richard Maltby's are freer and more challenging than most. Tough Cryptics has had its ups and downs on these questions, but it is now defunct. Emily Cox and Henry Rathvon are mellowing out, perhaps, though their disciples remain somewhat inflexible. The Enigma offers a healthy variety of cryptic constructors' voices and levels of difficulty, still largely within the US cryptic consensus. Trazom and I have succeeded Frank Lewis as cryptic crossword constructors for The Nation, and we are quite enjoying this opportunity to take a few steps away from tradition and find our own cryptic voice. Read more about this on Word Salad, our blog at

Outline of article:
The Problems
My Proposals
Time for a Change

I would like to start a dialogue on two problems I see in the growing US cryptic scene. Since the majority of cryptics published in the US appear in Games and Dell publications, their editors' views are well-known among constructors, and any alternative perspective must be defined in terms of theirs. Games and Dell puzzles are edited by Hex, WILLz, and Manx. I do not intend to deny or denigrate the enormously positive contribution these veterans have made to cryptics and to puzzling in general in the US. As a long-time cryptic solver and Games subscriber, I have much gratitude and respect for them, and will always be a fan. However I do not see eye-to-eye with them on all questions. I do not claim to know what's best for Games or Dell, only what I would like to see somewhere: Dell, Games, Tough Cryptics, or a hypothetical new cryptic publication. I am rather opinionated, but I hope this position paper will be seen in the spirit in which it is intended: not an attack, but a friendly challenge to the "cryptic establishment".

The Problems

1. There is a shortage of difficult cryptics.

After solving the excellent puzzles in Atlantic and Harper's, experienced solvers quickly run out of puzzles, and have to turn to British publications, or just do without. The cryptics in Games, for example, while appropriate to their audience, are just no fun to many four- and more-star solvers. The cleverness of some clues does not make up for the overall monotony of these cryptics as puzzles. The advanced solver never doubts that he or she will be able to finish the puzzle, and we have a mechanical solving process, not unlike that of the mass-produced daily paper puzzles: you just go back and forth between clues and grid until it's over. For advanced solvers, there is a need for more puzzles that are as hard as, or harder than Atlantic or Harper's but that do not require Chambers or extensive knowledge of Britishisms.

2. Narrow rules prevent the emergence of a variety of voices in US cryptics.

To my knowledge, only the Nation offers a non-square-dealing puzzle (for a definition of this technical term, see below). Frank Lewis is not my favorite constructor, but would the world be a better place if he did not exist? Absolutely not! In fact, there would be a greater pool of solvers and composers, and solvers would have more choices, if we had a wider range of puzzle styles, comparable to that available in the British press.

It is not necessarily true that beginners only like square-dealing puzzles. Approximately 17,000 people (17% of the readership of the Nation) attempt to solve Frank Lewis's puzzle every week. In my own experience of many years of group solving, I find that beginners can make more of a contribution to the group when we are working on Lewis's crossword, than on a normal American cryptic. This is because Lewis's chaotic style handicaps the experienced solver, but his creative way to look at the language offers many "ins" to people with varied interests and experiences. Over the last 40 years, many solvers have come to cryptics after having been introduced to them by Frank Lewis.

Note that Lewis plays within his own rules, which include having a few non-cryptic clues (usually puns) in each puzzle.

However, my main concern here is with the square-dealing camp, to which I belong, and more specifically with variety cryptics. The example that is usually given in response to the complaint of uniformity of style is the difference between the Harper's and Atlantic puzzles. I would add the example of Henry Hook, who certainly has a distinctive voice. But in fact Lewis, Cox and Rathvon, Galli and Maltby, and Hook are essentially unedited. Most of the puzzles that are edited (with the notable exception of those that appear in the Enigma) are extremely similar to each other.

If American-style crosswords were as uniform, it would be time for a new wave!



The appeal of a good cryptic crossword stems from the rich multidimensional web of word relationships and connections. This web has three components: the clues, the diagram, and the puzzle's gimmick (or theme). There is a lot of redundancy at each level. The clues provide two ways to get to the answer; the diagram offers one-half to two-thirds checked letters; and finally, the gimmick often provides additional information about some of the answers.

Cryptics could be too easy with this wealth of overlapping information. A good puzzle should seem nearly impossible at first, but provide enough ways to make progress to render the solution possible. At first, only some of the web is visible. As the solver advances, each additional part revealed adds to the total information available. The art of construction is to find ways to hinder the solver's progress without making it impossible. This can happen at all three levels:

Fairness requires balance. General square-dealing principles and diagram standards help provide a general framework for construction. If some information is taken from the solver by the gimmick in one place, more information has to be provided elsewhere. (For example, if some words are unclued, information is taken away, but if those words are all related by a common theme, information is given back.) If some clues are eccentric, others should be of a more familiar type. If an answer is obscure, its clue should not be.

Of course, there are solvers and puzzles at different levels of difficulty, and therefore there cannot be a single formula for the proper balance of challenge and fairness. That balance can only be reached in two ways: rules, and editorial judgment. Unfortunately, we have too many rules, and not enough flexibility in their implementation.


My Proposals

1. Creating more difficult puzzles:

Constructors and editors could occasionally heed these suggestions:

  1. Variety gimmicks could include modifications to the cluing/solving routine, instead of (or in addition to) the usual tricks that affect only the entering of the lights into the grid
  2. When the gimmick provides extra information, the cluing should provide less (e.g. if all answer words begin and end with the same letter as in the August Atlantic, the subsidiary part of the clue should only refer to that letter once.)
  3. Some puzzles could use obscure words as lights. Words should not be judged solely on obscurity. Some obscure words are boring, but others are interesting and fun to discover. Cryptics are great places to discover the latter, since if they are clued fairly and checked adequately, there are ways to get to them other than through an unfamiliar definition.
  4. Some puzzles could even use obscure words as parts.

In puzzles where obscure words are involved, it is preferable to have a standard reference dictionary. I realize obscure words are not everyone's cup of tea, and I am not suggesting all puzzles include them, but those solvers who enjoy encountering those once in a while currently have no place to find them but the Enigma.

2. Editing with a lighter touch:

This is probably more controversial. I would like to submit that the following cluing rule is the only one we need: the clue must read grammatically and correctly, at both surface and cryptic levels. This is the essence of "square dealing".

All other rules I see as aesthetic criteria, with which individual constructors and editors may agree or disagree, depending on their preferences, priorities, or mood, or even on the nature of the puzzle's theme or gimmick. For example:

  1. Etymological taboos in breaking up words and in double definitions
  2. Preference for conciseness over verbosity
  3. Rejection of superfluous words that do not affect the correctness of the clue (I do not propose accepting such words in hidden word containers, but they don't bother me elsewhere)
  4. Exclusion of far-fetched but poetic, bizarre, or interesting surface meanings
  5. Byzantine grammatical fixations
  6. Finicky connecting-word gospel
  7. Not accepting the full range of rebus-like readings (e.g. first person readings, or various somewhat forced readings)
  8. Singular/plural pickiness (which overlooks the fact that one word consists of several letters, and that one phrase consists of several words) leading to an unnecessary abundance of --ing words.
  9. Rejection of "flawed" and lit clues, e.g. ones where the whole clue is the subsidiary indication, and only part of it is the definition. This is especially objectionable because these clues are really fun, though admittedly not as impressive as perfect and lits.
  10. Rejection of playing with the spacing between words in the clues. (I do not advocate total heteronymy, except with adequate warning -- just occasional fooling with compound words. Why not "pigtail" to indicate "g"?)
  11. Rejection of triple definitions, and of "one and a half" definitions (what's so great about two?)
  12. Rejection of even mildly risque clues
  13. Belief that each clue must have a unique solution. (Redundancy makes this unnecessary -- remember, cryptics are crosswords!)
  14. Belief even that the subsidiary part of a clue must have a unique solution
  15. Picky anagram indicator restrictions
  16. Objections to using variety gimmicks in black-square diagrams
  17. Hard-to-fathom criteria for which abbreviations are acceptable. (U is OK for "school"?)
  18. Preference for whole-word anagrams over more complex clues, even if the result is a cliche
  19. Belief that every single clue must be solvable mechanically. (A clue is just that, a clue: "a piece of evidence that leads one towards the solution of a problem" -- MW. One or two clues per puzzle that are "enigmatic" in the sense that they lead to the answer in an idiosyncratic or devious way can add to the spice of solving, even at the expense of square dealing.)
  20. Rejection of the so-called "Cryptic Definition" clue (Alec Robins dedicates an entire chapter of his ABC of Crosswords, the square-dealer's bible, to these. See examples below.)

While I do not accept the above rules as rules, and notwithstanding my sometimes sarcastic tone, I do acknowledge most of them as valid aesthetic criteria. I certainly do not advocate that all of them be constantly violated. As a constructor, I try to stick by most of them, most of the time. (Though as a solver, I do enjoy some puzzles where they are ignored.)

No matter how valid these criteria are, there will be clues that break one or more of these standards, and yet are witty, fair, and fun. It's a shame to lose those. The end result of always excluding them is a monotony and predictability that benefits no one but the absolute beginner (and even that is debatable). On the other hand, allowing constructors more leeway while still adhering to broad square-dealing standards would liven things up, open the door to individual styles, and enrich all of us.


Here are some examples of supposedly faulty clues, together with their solution and the "rule" they break in parentheses:

From the Games Guide to Cryptic Crossword Creation:

  1. Man pronounced letters (4) (MAIL or MALE, breaks rule m. Since one of the last two letters must be checked, what's the problem?)
  2. Celebrity returned rodents (4) (STAR or RATS, m again. Two or three of the letters must be checked -- no prob.)
  3. Mets take in all clubs (6) (MALLETS, h. Reading: "The letters M, E, T, S take in ALL")
  4. Story includes Mom's Mexican food (6) (TAMALE, e. Reading: "The word described by the phrase "TALE includes MA" is "TAMALE")

From Harper's:

  1. Sneakily acquire whitefish? (6) (WANGLE, j.)

From Kornbluh/Picciotto:

(all of these were edited out from various puzzles)

  1. Someone seeking a mate around the university surmounted difficulties (4) (SWUM, c?, q)
  2. One can set this cam in order! (9) (MACHINIST, i. Interestingly, this was a favorite among our test solvers, but it was sacrificed to the rule.)
  3. Heartbroken clod (5) (EARTH, j. Was edited to: Heart-broken clod.)
  4. Undress halfway? Nudnik and I try, find no resistance (6) (NUDITY, l, r. Edited to: "The altogether untidy mess", which is a better clue, but transparent to the experienced solver, as the anagram is an old standby.)
  5. Old story: "One who is crazy about breasts" (8) (CHESTNUT, l. "... breasts" was changed to "strongboxes", weakening the surface meaning.)
  6. Card enters pack hot (6) (ARDENT, h? Reading: "the words CARD and ENTERS pack -- =possess, 9C -- the word ARDENT. This clue is concise, tight, and a difficult hidden word clue, one of my favorite goals as a constructor. Plus, it included my nom. Oh well.)
  7. Pole to box mineral (4) (SPAR, k)

From The ABC of Crosswords:

  1. A wicked thing (6) (CANDLE, t)
  2. Its pages are used for drawing on (9) (CHECKBOOK, t)
  3. They arrive after the milk and stay for countless meals (9,5) (PERMANENT TEETH, t)
  4. A stiff examination (4-6) (POST-MORTEM, t)
  5. A number of persons (11) (ANAESTHETIC, t)

(Actually, I would never write clues of this type, but I do enjoy solving them.)

From a list of all-time favorite clues among some InterNet users:

  1. Swell dance in the nude (7) (BALLOON, s. From the London Times.)
  2. Unhappy? Go chew a pillow! (4,2,3,5) (DOWN IN THE MOUTH, s)
  3. Might such an unsophisticated wordplay be said to amuse John H. Watson, MD? (8) (HOMESPUN, s)
  4. Mantle of madness (6,7) (MENTAL DISORDER, s. From the Manchester Guardian. This type of clue, including a sort of cryptic indication in reverse, is used a lot by Frank Lewis.)
  5. (8) (CLUELESS, s)

In addition, I have included with this position paper a recent Harth puzzle from the Enigma (X3 in the February issue.) The puzzle includes a number of violations of the above rules, and yet Treesong and Sibyl published it as is, without changing a comma. Well, the sky did not fall, nor was the puzzle highly unpopular. Many very sophisticated solvers seemed to enjoy it: almost half its solvers awarded kudos to it, and I believe it got the most bangs in that issue. (See February stats in the May issue.) Were these solvers not smart enough to see the flaws? More likely they did not care. (Rule violations: 8,o; 14,c; 16,o; 18,b; 19,o; 20, g; 22,i -- and yet rated "excellent" by Treesong; 23,n; 32 must violate something, but I don't know what. Also 15,a, which could have been avoided with a first sound change on "selfish", but that didn't occur to us at the time. The whole puzzle violates p.)

This is not the best cryptic ever written. But it's a decent puzzle, and it was fun for some experienced solvers. Perhaps some editing could have made it better, but the sledgehammer editing of most US cryptic editors would have tended to take away its personality. Similarly, the above sample clues are not the best clues in the universe, but many of them would be fun to come across while solving a puzzle.



Working around constraints is to a great extent what clue-writing is about. Hex are the geniuses of the concise clue. (Of course, far more important than their conciseness is their wit! On occasion, they add even more constraints: in one recent Atlantic puzzle, every single clue had a reference to playing cards.) Fraz likes to avoid connecting words. Treesong, as an editor, has high standards for surface meaning, another sort of constraint.

And on it goes. Many constructors and editors have pet constraints, and it's no accident: constraints can be a tremendous help to creativity. The list above is one possible set of constraints, but by no means the only possible list. For example, Trazom once pointed out to me that it is in poor taste to use made-up titles in clues. If he wrote more cryptics, this constraint would probably be on his list. Another legitimate constraint is to avoid reusing the same abbreviation, or anagram indicator, or other device within the same puzzle.

Personally, I like to vary my constraints from puzzle to puzzle. In one puzzle I tried to include pairs of opposite words in as many clues as possible, which led to some good finds (e.g. "Suggest writing half of poem in prose (7)"). In another puzzle, I decided to clue every single theme word with a "Just the Opposite" type clue, (as a satire of this bizarre cluing convention.) Clearly, I have no objection to constraints, even severe constraints. But we don't need to all have the same ones!

Finally, when the puzzle's gimmick includes additional constraints of its own, (as often happens with our puzzles), something else has got to give somewhere. For example, if many words must be placed into the diagram, symmetry may have to go. If the clues are forced to include a certain letter, then some other constraint, such as the quality of the surface meaning, may need to be relaxed.

Constraints can backfire, and be a corset that stifles constructors. At the other extreme, they can be underused, leading to chaotic, unfair, and unsatisfying puzzles. Paradoxically, it is sometimes in the judicious violation of tight constraints that the most creative work happens. Someone once wrote to GOTS about enjoying a Harth puzzle in spite of its flaws. Perhaps in reality it is those very flaws that made the puzzle fun! And perhaps that solver would have had even more fun by relaxing a little. In fact I suspect many constructors and editors ruin their own enjoyment as solvers when they cannot let go of their fixation on the list of so-called rules. It is perfectly possible to enjoy a puzzle that is written in a different style from one's own. (For example, take Beacon's Alice set of puzzles from the last convention. It was great fun, even though one could easily complain about this or that technicality. Relax, people! Last I checked, cryptics were a recreational activity!)


Time for a Change

It may be that over-emphasis on rules developed at a time where cryptics were a novelty, and constructors needed clear and strict guidelines. It may be that this approach is still appropriate for Games and Dell solvers. But times have changed: there are more constructors, who know more about cryptics, and there are more solvers, some of whom want more variety or more challenge.

I for one, (I cannot speak for Arachne), would never have been a constructor if it had not been for Treesong's relative liberalism as an editor. Similarly, Mr. E could only have emerged in the Enigma. (It is interesting to note that both Harth and Mr. E started out anonymously.) Treesong understands the "rules" as well as anyone, but he is able to overlook them when appropriate. Whether you, dear reader, like our puzzles or not, some people do like them. Those who think we are too loose are under no obligation to solve our puzzles, but I fail to see why things would be better for people who do enjoy our puzzles if we had no place to publish them.

There is no risk of traditional American-style cryptics disappearing. Hex, Wabbit, Fraz, and many others will continue to write in that style. But as new constructors enter the field, let them find their own voice within flexibly-enforced square-dealing standards. We do not need more clones of the constructors we already have. It is time to move from "Everything that is not allowed is forbidden" to "Everything that is not forbidden is allowed." Better yet: "Think twice before forbidding something!" Quality cannot be judged merely by adherence to rules. Instead, editorial judgment must involve a balance of aesthetics and fairness, and discipline must be weighed against creativity. Let cryptic constructors develop some personality, and let the solvers decide what they like!

If this approach is not appropriate for Games or Dell, perhaps some of the other publications could open up a little? Or will the Enigma remain the sole outlet for variety and challenge?

(PS: I started writing this before Tough Cryptics. Perhaps this new publication will grow into the answer to my prayers.)



This is a follow-up to the above article, written for Graffiti on the Sphinx (a publication about The Enigma) in December 92, and very slightly edited:

In addition to the comments in GOTS, I received several letters and spoken comments from constructors and editors in response to my cryptic manifesto. A range of views were expressed, particularly in reference to various items in the list of "rules", and to the objectionable clues I defended. Few agreed with everything I said. I received very positive comments from many, and I also got vigorous disagreement -- most people were somewhere in between. Fans of British puzzles tended to be friendlier to my ideas, people associated with Games less so.

I am not sure whether Qaqaq (9.11.11) implied that I was against square-dealing. I certainly am not. Willz (at the convention) had a much more accurate appraisal of my position: he complained that I was too close to the dominant US square dealing standards. (His point was that if I am this close, why not just accept the prevailing opinions.) I would agree with Willz that what I call for is not fundamentally different from what we've got. I just want some loosening in order to open the door to harder and more varied puzzles.

I strongly agree with Qaqaq on restrictions. As he says, the artistry lies in working within constraints. I was surprised that he did not seem to notice that this was exactly the point of much of my article, not to mention of many Harth puzzles. No one could argue that we (Harth) were not working within exceptionally severe restrictions when constructing "Diagnose!" (December Enigma) or "Alack!" (September Tough Cryptics). To repeat my position: constraints are what it's all about, but we don't all need to work within the same ones. And each of us need not work within the same ones every single time. Finally, breaking constraints is sometimes worth it, but only if it is done deliberately, knowingly, and for good reason. (It is probably true that my tendency is to err on the side of pushing the limits too often, an over-reaction to the very uptight environment of US cryptics.) Variety is the spice of puzzling, and no one knows this better than boundary-stretching rebus genius Qaqaq.

I am curious about what a solver of Qaqaq's caliber looks for in a cryptic, since the puzzles he defends are boringly easy even for me, and I am a much worse solver than he is. If he enjoys solving cryptics, where does he find puzzles worthy of his talents? Perhaps we have a different idea of what the point of solving is? Or perhaps he doesn't enjoy solving cryptics, just constructing them?

His claim that looser editing would turn off beginners is disproved for example by the popularity of Alan Richardson among non-experts. (See Wabbit's comments, 9.11.12.)

100 Down (9.12.13) asked what I mean by a "one and a half" definition. It is a double definition where the two parts have a substantial overlap in meaning, which makes the clue be less than double. This is often a fatal flaw, but it does happen once in a while that a great surface is obtained that way, or that a 1.5 def clue is pleasant for some other reason. Again, this is not something I advocate as policy, just a transgression that is sometimes defensible. (See example below.)

I take special delight in Treesong's reaction to that rule-violating clue of mine. He rated it "excellent" in a moment of inattention, which must indicate that he enjoyed it when solving the puzzle. Then he withdrew his praise (9.11.11) when I pointed out that there was a violation! I find this an oddly puritanical attitude. If the clue seemed excellent, perhaps that's all it needs in order to be excellent! I wonder how often solvers enjoy the illicit pleasure of a witty but sinful clue that got past an editor. And what sort of a system we have that requires editors to sleep on the job for a good but unorthodox clue to get through. For example, Henry Hook's clever "Jagger's garbage is nearby (1, 6, 5, 4)", brazenly violated the etymological taboo in a recent Tough Cryptics. I guess that makes it a less-than-double def. Did Famulus allow it by mistake? or was it deliberate? Is a violation acceptable if the author is established enough? is it enough to have a solvable and witty clue, and damn the "rules"?

I will not get into a discussion about specific "rules" here. My point is precisely that if we agree on my one all-embracing rule (meaningful surface and cryptic readings), there is room for differences on specifics. Again, I respect many of the so-called rules I listed. My point is not that they should be violated randomly, just that they are not God-given. (While Hex and Manx are certainly god-like in my view, they are not God!) It is clear from the range of responses I got to my article that you would not be able to find agreement among U.S. square dealers as to which items belong on a list of rules, and which do not. There are already some differences between Games, Tough Cryptics, and Enigma. If more publications come into being, there will inevitably be more variation, and even these three will diverge. It's just a matter of time. My article was intended to start some discussion, which it did to a limited extent. Those who think there is nothing to discuss should open their minds, if only a little, and hear the range of views among constructors.

Visit Hot's Puzzle Page. Send me e-mail .