World’s First Ambigram Card Game Released

Jul 10th, 2009 | By | Category: Products
Ambigram, the game

Ambigram, the game

FlipScript has released the world’s first ambigram card game, called simply “Ambigram”. The game is notable in that all of the 50 playing cards display an ambigram on both the front and the back of each card.

The rules are simple: Each card in your hand will contain an ambigram, displaying one word when it is read upright, and another word when it is read upside-down (like “Angel” / “Finger”).  You can play either of the two words in your ambigram on either of the two words that are currently displayed on the deck (like “Drink” / “Angel”).

The first few rounds are simple, since you will have lots of different ways to play the possibilities in your hand, but you had better plan ahead and pay attention not just to your own cards, but to what cards your opponents have, because the game gets a lot trickier in the later rounds.  You don’t want to get caught drawing a fist full of cards when your opponents are almost out!

To make the game even more challenging, your opponent might zing you at any time by slapping you with a red “Double Joker” or a red “Double Angel” card that contains the same word in both directions, doubling the chances that you’ll have to draw!

Of course, you could plan to zing them back with a red “Double Siren” card, but don’t get caught holding that card in your hand for long, because not only is it twice as hard for your opponent to play on, it’s also twice as hard for you to get rid of!  You had better plan ahead, or your scheme to attack at the end when it can hurt the most might come back to bite YOU!

Simple to learn, but filled with plenty of twists (literally), Ambigram! the game provides hours of fun for the whole family.

About the Playing Cards

The Ambigram deck itself seems impossible at first, and in fact, it would have been impossible just a few years ago. No human being could have designed it alone, even if they possessed several lifetimes to work on the problem.

For example, below is a picture of one “suit” from the game: the complete set of 7 “Joker” cards (there are also 7 such suits in the deck, for a total of 49 ambigram cards).  You will see that all of these cards say “Joker” when presented in this orientation, but all 7 of them also display a different word when turned around.  Every single card in a “suit” is totally different, yet viewed the other way, they are also all the same.

The center card has its corner ambigrams displayed in red since it says “Joker” in  both orientations.  The red cards become a critical part of effective game strategy.

One "Suit" from the Game

The Joker "Suit"

To find exactly what words to use for the deck, a computer was programmed to go through the entire English Dictionary, word by word, and locate 7 words that could all form an ambigram with themselves, and also form an ambigram with every one of the other 7 words that were being located – a total of 49 ambigrams in all.

The complete Ambigram card deck (click for full view)

The Ambigram card suits (click for full view)

It took the computer 16 days to go through the Dictionary and return the answer: Angel, Drink, Finger, Joker, Killer, Learn and Siren - the amazing set of words that make up the 49 cards for Ambigram! the Game.  Those 49 ambigram cards are joined by a single green Wild card, for a total of 50 spinnable round cards that play out differently every time!

Ambigram! the Game is available from FlipScript Ambigram Products for $18.99, although introductory pricing is currently in effect.

Tags: , , ,

Leave a comment »

  1. This sounds like a neat game for ambigram fanatics. From the general description, it seems like it is played in a manner similar to Uno, where you must play a card that matches the exposed card ‘in one way or another.’ I’d love to try this game…. if only I knew someone else who would be ‘crazy’ enough to play it with me.

  2. Hi Rebecca,

    Yes, you are right. The game play is similar to Uno, with the goal being to rid yourself of cards, and to play on the exposed card in “one way or another”. The major difference is that there is really only one way to match the exposed card in “ambigram” (the words need to read the same), whereas in Uno you can match either color or rank.

    So, if you happen to be both an ambigram lover *and* an Uno lover, you should definitely try it out.

  3. “And in fact, it would have been impossible just a few years ago. No human being could have designed it alone, even if they possessed several lifetimes to work on the problem.”

    What an crock. Somebody give me a list of any eight other five-to-six-letter words. I will get to throw out one (which will be my counterpart of having a computer scour the entire English language for words easy enough for Flipscript). I’ll see how long it takes me.

  4. “Human vs. Machine”

    I like it.

    OK, I made it a little easier on you by picking out 8 words that all contain 5 letters (there are no six letter words). I just grabbed them from a list of valid 5 letter Scrabble words (the list is here: and picked out some that all start with a different letter (as was one of the conditions for the cards) and are words that most people would know.

    As you requested, the list contains 8 words, and you can throw away any one of them so you have 7 words remaining. The challenge will be to create legible ambigrams for:

    * Every word to itself
    * And every word to every other word

    Here is the list:

    * doors
    * enemy
    * icing
    * libra
    * momma
    * olive
    * squat
    * works

    If you pull this off, not only will we send you a free copy of the game, but we’ll write up a entire article about your accomplishment and post it on the home page for all to see (along with our necessary crow-eating of course).

    Good luck!

  5. Looks like Kevin has come up to scratch:

  6. Hey Kevin,

    Not bad! To be honest, I really didn’t think you’d take up the challenge, so I have to give you props for giving it an attempt.

    Unfortunately, a lot of the ambigrams are just not legible. Heck, you actually could have used the “ambimatic” to create so-called “ambigrams” of these words since they all have the same number of letters, but they definitely wouldn’t be ready for a commercial game.

    The word selection algorithm I mentioned looked for 7 words that would all design highly legible ambigrams (in the terminology of the program, every ambigram needed to have a score of 90 or higher on the legibility scale… a lot of yours would have scored about a 40. At best.)

    Here are just a few that I couldn’t read (and what they look like to me):

    Keep in mind that just a couple of illegible designs would completely destroy game play.

    I appreciate the work you put into this, and I think that it provides further evidence that optimal word selection is paramount in trying to create a set of 49 ambigrams that are all highly legible in both orientations.

  7. We’re not welching on the bet, though!

    If you want a copy of the game, just send us your postal address (privately), and we’ll have one shipped out.


  8. I wonder how much of it is related to what we’re used to! I’ll be honest, it’s hard for me to read your provided “Joker” suit- I have an easier time with Kevin’s set. (Although I agree on “Savot”, and “Skjora” until my brain realizes that K does not point that way. I had no problem with the other two.)

    But I’m used to Kevin’s style of ambigrams- while ambigrams derived from blackletter typefaces are something I see less of. You’re used to reading blackletter ambigrams, so they’re easy for you to process; I’m used to reading Kevin’s more Uncial-esque style, and so it’s easier for me. Of course your computer would hate Kevin’s work; it looks nothing like the modified blackletter it uses to identify a good match!

  9. I agree with Windrider above–I find Kevin’s ambigrams, for the most part, much more readable than the ones shown from the card game. Sure, some of his are hard to make out unless you know what you’re looking for, but the same is true of the ones shown from the game–even knowing the word list, it was very hard for me to see what word some of the cards in the Joker set were supposed to be when inverted, and some of them were so completely illegible I pretty much had to figure them out by process of elimination; there are already “a couple of illegible designs” just among the cards shown in the pictures! (They may score 90 or higher by the computer’s legibility scale, but they’re not remotely legible to me; I don’t know what algorithm the computer uses to measure legibility, but it doesn’t seem to correlate very well with legibility to humans, or at least not to this particular human.)

    Which isn’t intended to put down the game; it’s an interesting idea, and trying to make seven words all ambigrams of themselves and each other is definitely a challenge. But criticizing Kevin for including ” a couple of illegible designs” seems very much like a pot-kettle-black scenario. The ambigrams shown from the game are very, very far from being “all highly legible in both orientations”; if the cards shown from the pictures in the article are typical, Kevin’s done at least as well using the arbitrary set of words provided as was done in the game for the computer-chosen set of words.

  10. More specifically, since you gave specifics about what some of Kevin’s looked like to you, here’s my impression of the pictured Joker suit:

    It’s easy to tell, of course, the double Joker; of the rest, the best is probably Siren, but even that one I wouldn’t characterize as “highly legible”. The first letter doesn’t really bear much resemblance to an S, and the middle letter doesn’t look much like an R, but they look even less like any other letters, so it’s more or less possible to puzzle out what it’s supposed to be. (Actually, the middle letter looks to me a little more like a G than an R, and the last letter looks as much like an R as an N, but Sigen, Siger, and Sirer aren’t words, so…) If I hadn’t known the word list, Killer looks more like Hiker (yes, the K would be backwards, but that’s what it parses as to me–the K is backwards in the “Skjora” in your post, too), Learn like Lager, and Drink like Arion or Agon. Angel is unidentifiable, except by process of elimination… that first letter really looks much more like a D than an A, and the second letter doesn’t look like much of anything, though once I know what it’s supposed to be I can sort of see it. Finger I can’t even say that about; even after knowing by process of elimination which one it must be, I _still_ don’t see anything there that looks like an F. (I mean, obviously I can tell which part of the ambigram is _supposed_ to be the F, but it doesn’t look anything like an F to me.)

    Now, different people have different perceptions, and maybe some people have a much easier time with these ambigrams than I do… but still, as I said, to me, Kevin’s overall are much easier to read; for me, at least, the proportion of designs that are “just not legible” is lower among his than among those pictured here. So, no, I would definitely not agree that this exercise “provides further evidence that optimal word selection is paramount “–exactly the opposite. I don’t think your words are nearly as legible to people who aren’t intimately familiar with your specific style of ambigram as you apparently think they are…

  11. The computer decides how legible it is, too? Ha, oh dear. At least these misreadings aren’t common English words like, say, “Anger”. With a few more “lifetimes” I imagine I could work out the bugs (My r’s don’t absolutely need to reach past the baseline, for example) , but I do have other things to do.

  12. Wow, Kevin – Looks like we’ve hit a nerve here. There definitely seem to be some strong opinions about this.

    I will state again that I’m impressed with what you’ve done. You took the challenge and ran with it. I give you props for knocking this out. However, until the libra/works and squat/olive designs (and others) are at least partly legible, I maintain that word selection is the key to success in this.

    In fact, I’ll bet if you did the same challenge you just completed, but used the 7 words from the game instead, you would create some designs in your “Uncial” style that absolutely ROCK. The point of that section was not about the “style” of the letters (which is what everyone seems to be focusing on), but about the word selection. Finding 7 words (with the conditions that were placed on them) that all make legible ambigrams to each other is the key to making a nice set… in any style.

    I’ll state a little bit of the obvious here, because it seems like it may have been missed. Your image was a good size table of designs, whereas that “Joker” suit is a teeny-tiny, compressed JPG. That image was just a little graphic to go along with the text for the article, and wasn’t intended to be downloaded, photoshopped and rotated in an attempt to read the inverse words.

    The point of the image, of course, is to show that the same word can invert to several other words. Take a look at it, and see if you can read the word “Joker” on each card? Yes? Great. No? OK, which card are you having problems with reading? They all look like “Joker” to me.

    I can post larger versions of the designs, or post another “suit”, but like yourself, I also have other things to do.

    Yes, the “Joker” cards are similar, and that is for a reason. When you have the deck in hand, you can instantly read every card. Yes, knowing what the words are supposed to be ahead of time certainly helps, but so does the consistency in the design style. They are a “set”, so this is all by design.

    By the way, a human determines the scoring. The computer just does the addition.

  13. Agreed with the others — I’m reading this on a teeny smartphone screen, which reduced Kevin Pease’s design to the same size or smaller than the Joker set on my view. And yes, I had to struggle to read some of his ambigrams. But of the ambigram game’s samples, I can’t make out three of the seven inversions (they’re all legible right-side up as “joker”, but upside down I can only pick out “angel”, “joker”, “killer”, and “vigor”.

    … errr, whoops, vigor isn’t on your list of words. Sorry. So I’m only getting three. I still can’t puzzle out finger, drink, learn, or siren, even by process of elimination. :/ Maybe I’d have an easier time at full size, as you say, but Kevin’s is just as small on my little screen.

    Anyway, nice job to both of you regardless — it’s quite a challenge, inverting all those words to match each other.

  14. If you’re trying to show how one word can be reversed to several others, it seems to me you’d want a higher-quality image, so you can demonstrate exactly that! In the context of “joker joker joker joker joker joker joker”, it’s not unreasonable to read all of those cards as “joker”, and I agree that your style has a lot more visual unification than Kevin’s. (Olive/Doors and Olive/Enemy are not stylistically unified!) But if I blind myself to the other six Jokers on the board, the second one can much more easily be read as “Yokel”, the third and fifth are both “Jager”, and for some reason my eye wants to parse the last one as “Jakel”- which is completely unfair to you, since that reading has to ignore the rightmost line of the shape. Can they all be read as “Joker”? Yes, given the context of the restricted word list. But given that context, the only one of Kevin’s (Among those you pointed out) I can’t read is “Squat/Olive” (Savot), and Kevin’s a creative fellow- I have no doubt he can revise that one.

    But without the word list, I’d honestly have trouble reading the second of your Jokers as anything but “Yokel”. I know this is a low-res version, but I flipped the image over in Paint to experiment anyway- to see the alternative words, with as much resolution as they had when presented as “Joker”. ‘Killer” and “Siren” are good. “Finger” reads as “anger”- and, with the word list, I assumed it was actually “angel” until I reached the actual “angel”, which is quite clear as such, credit for that one! But I can’t read the next two at all. Without the word list, I get “prior” and “lener”, sort of, for the last one, maybe another “jager”. To be fair to you, “learn”- which I could identify with the word list and effort, I still can’t see “Drink” as such at all- definitely looks like it would improve a lot at a higher resolution.

    Your product is interesting and unique- but you’ve got a whole community to leverage! Why not give a pre-release preview of the product and ask for feedback on which ones you should redesign? (I don’t think you have anything to lose by completely redoing Joker/Drink, I can’t read it either way.) It’s the quality and styling of the cards as an art piece that is their unique selling point, and showing them to people should not make them want it less. I certainly don’t think people printing up their own decks is that big of a threat- you either get low-quality home inkjet dreck, which defeats the purpose of an art piece, or production significantly more expensive than just buying it from you only to get results that are still low-quality relative to print standards (based off even a good resolution for the ‘Web).

    I think both you and Kevin have good points. You’re right that word selection is important (‘Works” seems to have been especially tricky- given that I couldn’t read “Drink”, maybe “K” is just bad), but I think Kevin’s right that it really didn’t require a computer to do it- just someone with the idea (which you had, and deserve credit for), and a few days to experiment.

    I can see why Kevin dropped “momma”- there isn’t enough variability in letters for him to work with, since the same letter needs to look pretty much like the same letter through a word. If he was taking on this project from the outset, and happened to start with that remaining seven-word list- well, I’ll just ask. Kevin, which word would you switch out for what to get the next iteration of the experiment?

    …actually, using a computer sounds like it would suck a lot of the fun out of the experimentation and exploration to find out just what you can do.

  15. Take a look at it, and see if you can read the word “Joker” on each card? Yes? Great. No? OK, which card are you having problems with reading? They all look like “Joker” to me.

    So the fact that some of the inverted words are completely illegible is okay as long as the “Joker”s are all readable? Again, the “angel” and “finger” in the inversions are so totally illegible to me that even knowing the word list I had to figure out which words they were by process of elimination; that’s already “a couple of illegible designs” right there. (And frankly, no, the Jokers aren’t that great either, though they’re better than the reversals; the fifth one looks more like “Jaked”, and the sixth like “Jakeg” or “Jakig”. The rest I probably could have identified as “Joker” without knowing beforehand, but it would have taken some work.) The bit about the compressed JPG is a very weak excuse; the outlines of the letters are still visible just fine, and I really doubt they’d be any more readable in a larger size. (For the record, rotating an image 180 degrees in Photoshop does not in any way affect the image quality. Rotating by other angles, yes, but there’s absolutely no degradation in quality for a 180 degree rotation.)

    By the way, a human determines the scoring. The computer just does the addition.

    I don’t see how that could possibly work, unless the computer does some preliminary scoring, and you’re just saying the human scoring comes in at the end to check its work. Even if the computer starts with a set of just 100 words to winnow down to its final 7, that’s still 5,050 ambigrams to be scored. And if you started with, say, 500, that’s 125,520 ambigrams. Surely you don’t mean to say that a human scored each of those ambigrams individually. Or if you’re saying that a human determines the scoring not of the ambigrams as wholes, but of the individual letter reversals, that creates other issues… symbols may parse very differently depending on what other symbols they’re next to, so just putting together letter reversals that you find individually legible won’t necessarily result in a legible ambigram. (Though many of the individual letter reversals I don’t find legible, anyway… as I said, the F of finger is totally unidentifiable to me even after I know what and where it’s supposed to be, and to me ironically the symbol you used for an A in Angel looks more like a D, and the symbol you used for D in Drink looks much more like an A…)

    Also, is just one human doing the scoring for each ambigram (or letter reversal)? Because (as we can see here) different people have an easier time with some ambigrams than others, so you may end up with a set that’s optimized for one particular person’s eyes but is unreadable by anyone else. Especially if that person is used to seeing ambigrams done with these specific letter reversals, and knows what to expect.

    Anyway, I suppose the details of the scoring algorithm don’t matter given the end result, which is that despite all your protests it clearly just didn’t work. However much you make excuses and repeat that they’re all perfectly legible, most of them are not at all legible to me. Sorry, but I still think you’re being influenced by knowing in advance what your words are, and knowing the specific letter reversals you’re used to using. Your computer may have selected seven words that score high according to some particular algorithm, but when it comes to actual human readability I think Kevin’s demonstrated conclusively that this can be done just as well with an arbitrary set of words.

  16. I will say this, however (and I apologize if my previous post came across as a little too negative; I should have tempered my criticism more): you’re right that your set does have the advantage of stylistic unity. Yours do indeed all at least have the same “look”, which Kevin’s don’t. His seven Libras, for example, look like they belong with completely different fonts; your cards all have a much more unified feel. So definite points for you there. And I don’t mean to minimize what you’ve done; coming up with all these ambigrams was certainly an accomplishment. I just think you’re seriously overestimating their legibility to people who aren’t already very familiar with this specific style of ambigram, and placing far too much stock in the computer’s word selection… and you’re way too defensive when you’re called out on it. I just think your claims that it “would have been impossible just a few years ago” and “No human being could have designed it alone” are rather overblown, but still, it’s an interesting idea, and I don’t mean to be too negative about what you’ve done.

  17. My friends can be generous with their opinions, and I hope it doesn’t look like I’m leading an invasion. I consider demonstrated everything that can be demonstrated. I did what I wanted to do, and you have my address, so you can send the game if you wish to; you certainly didn’t have to offer it considering that I put up no stake of my own.

    I may try it again sometime with a thematic set of words. For now, though, there are more compelling projects at hand.

  18. It took me a couple read-throughs to see that the green card is supposed to say “Wild.” It’s been noted– okay, Kevin Pease has noted that people tend to read words along the top edge, and the triple-serif makes it look like “Mild.”
    And I agree that the Joker images are too small. If you arranged them in rows of 2-3-2, you could more than double their apparent size without sacrificing horizontal space.

  19. Interesting commentary. I actually bought the game for my autistic daughter– she is obsessed with card games. She had a little difficulty reading a couple of the words the first few times we played, but she had little difficulty reading the majority of them. She loves this game, and I find it challenging. It is a lot like UNO– her favorite game prior to this one. As far as I am concerned, ambigrams, in general, are a form of artistry where legibility is truly in the eyes of the beholder. In my opinion, the ones posted by the ‘competitor’ were barely legible in most of the orientations. I even showed the words to my daughter and, not realizing that the words were the same across each row, she could only identify 1 word in each row– except for enemy, where she identified 2.

  20. I am really enjoying this. This is by far the most spirited debate on the site.

    If nothing else, I think that the debate shows that beauty (and legibility) are truly in the eye of the beholder. Its also possible that we become intimately familiar with what we repeatedly experience (just like learning the character set from our first language), and everything else seems foreign.

    Just as I can not wrap my head around some of these comments, I have to accept that it may be equally true that others may not be able to see things as I (and other familiar with our designs) do.

    What we *really* need are some people that have seen neither these blackletter designs, nor Kevin’s designs, to weigh in with an opinion.

  21. Start submitting this post as possible links of interest to other typography blogs, or general nerd-interest blogs; maybe you’ll get a larger audience? I already asked BoingBoing if they wanted to weigh in.

    …oh, “wild”! I thought the green card said “win” until I realized there’s an extra arm there, and then I couldn’t figure it out. Although if you use it at the right time, maybe there’s not so much difference!

  22. I was thinking that it’d be nice to print one line showing all seven words from each set at the same resolution, then ask respondents to post what words they read from it. That avoids the bias of reading the same word each time once you see the pattern they’re in. Kevin’s designs were easier for me to read right-side up than upside down (reading on a smartphone, I just flip the phone over) because I didn’t recognize the pattern in the columns at first, but I did pick up the pattern in rows.

  23. By popular demand, I have updated the original article (above) with a link that shows all 7 suits of the deck.

    You can also view it here.

    If it comes up automatically “resized” in your browser, be sure to view it at full resolution.

  24. I still think you’ve missed the point. The question isn’t whether Kevin’s ambigrams are better than Flipscript’s; the question is whether your original claim–that no human being could have done this over the course of several *lifetimes*–was reasonable or not. And given that Kevin took a tiny fraction of a lifetime to accomplish a similar task, I’d say your claim has been proved ridiculous if his ambigrams are even a quarter as good.

  25. I don’t want to take away from the great work that Kevin did here, but the original claim was not at all “ridiculous”, and in fact, still stands.

    The claim was that it would take a human an exceptionally long time to find the perfect 7 words for a 7×7 ambigram grid. The claim was about WORD SELECTION. Feel free to scroll back up and read it.

    Most ambigram artists (including Kevin) would agree that word selection is very important in creating even a “simple” 2 word ambigram (one word to another). Add a third word to the original two and it becomes MUCH, MUCH harder. I would be surprised if a human could even do the word selection for a 5 word grid, much less 7.

    Kevin’s example using 7 random words actually illustrates the problem perfectly.

    From the comments, it seems that most people would agree that the SQUAT / OLIVE design is not readable. It looks like SAVOT (although putting letters on top of each other really should invalidate the design completely… normal words have one letter after another). Anyway, for argument’s sake, let’s pretend that SAVOT is the ONLY illegible ambigram in the set. So, either SQUAT or OLIVE has to be thrown out from the developing word set. Let’s replace SQUAT.

    We now need a word that can make a legible ambigram with all 6 of the remaining words (and itself). Invariably, it will work with some, and not others, forcing us to drop one or more of the remaining words from the list and replace them as well.

    After a few replacements, you may find that replacing SQUAT did not help. So, we go back to the original set and replace OLIVE. As you can guess, the replacement branches multiply exponentially. This process would continue on, and on, and on…for months, years, or yes, my guess would be a lifetime.

    I suppose you could “cheat” and choose 7 words that are composed of just the letters “i”, “t” and “l” (purely vertical), or “o”, “a” and “e” (purely circular), or that have the same basic repeating character patterns, but minimizing letter variation would make every ambigram look too similar to allow them to be easily distinguished.

    Not only that, but even if it were possible for a human to somehow miraculously identify a set of 7 words with normal letter variations that could be found to make a respectable 7×7 ambigram grid, there may be OTHER such sets that would create even BETTER ambigram grids. The goal here was to find the OPTIMAL set of 7 words to make a 7×7 ambigram grid.

    Good luck.

    The challenge was never about creating something that could be called an ambigram. As I mentioned, the “ambimatic” could have completed THAT challenge with the same words given to Kevin in about 28 seconds… one second for each unique ambigram.

    The original claim was that a human could not have done the WORD SELECTION to create the 7×7 grid. The actual drawing of the designs is almost trivial compared to that challenge.

  26. Word SELECTION? Ah, nuts. We’re never going to get to the bottom of this.
    Two other things. First, the article says that the DECK would have been impossible a few years ago, and doesn’t specify which is the “impossible” part, the selection or the ambigramming. Second, I can argue that the CURRENT word choice is not optimal, since a couple instances of “Finger” look like “Anger,” or even “Angel.”

  27. Actually, the article has always specified what was the difficult part.

    Read the paragraphs starting with “To find exactly what words to use” and “It took the computer…”. They are right there in the “About the Playing Cards” section right after where it states that the the deck would not have been possible a few years ago.

    Of course, creating 28 or 49 ambigrams WAS possible a few years ago. The ambimatic has actually been able to do that for more than a decade, but it sure doesn’t produce designs that are readable, or are stylistically similar. The difficulty has always been the more difficult challenge of word selection.

    Regarding your second claim that the current word choice is not optimal: You are actually correct. There were 3 other sets of 7 words that were “better”, but they all had 1 undesirable word in the set of 7 (either profane, unknown or bizarre in this context). This list is actually the 4th best set of words for creating a 7×7 ambigram grid, but it was the “best” for creating a commercial game.

    In fact, I almost threw out this list, too, due to the word “killer”, but lists 5, 6 and 7 were no better, and I would have gone too far down in quality to get a set to use, so I kept the list that contained “killer”.

  28. By the way, which “finger” looks like “angel”?

  29. Short answer: probably the Finger Finger.
    Long answer: here’s a table of the questionable cards. Each period represents a word that was fine, while each question mark represents a word I might have misread out of context, with some possible misreadings in order on the right. (Best viewed in fixed-width font.)

    .?.?..? Yarn, Leabn, Leary
    ?.???.. Aeink, Arink, Ariok, Arink
    ……. (none)
    ?.?.?.? Jgker, Jokgr, Jokeb, Jakey
    .?.???? Qnger, Onger, Fingel, Finges, Anger
    ..????? Firen, Siker, Siged, Sing, Sirey
    .?.???? Killid, Kiyer, Kijlen, Hitler, Killey

    Of course, I admit I probably underrated my field performance and overestimated the rate of mistakes here.

  30. You know, Cy, I think that about wraps it up for me. You are clearly no longer providing anything even close to an unbiased opinion, and this discussion has grown tiring.

    I definitely don’t see what you’re seeing.

    The first “joker” card says “Jgker”? Honestly? That second letter is a “g”?


    The last “joker” says “Jakey”?

    How do you see an “a” as the second letter there.

    Are you *trying* to misread these?

    And your “Fingel” and “Anger” mis-readings both start with the same character! It can’t be both.

    That’s only halfway through your list, but I’ll stop there. I’ve seen enough.

    Anyway, I’m done.

    I have stated my theory, and until it is disproven, I stand behind it. Here is the simple form of the theory:

    “A computer can do word selection for a complex 7×7 ambigram grid more efficiently than a human.”

  31. I’m sorry. Yes, I was trying to misread some of them, and I will try not to defend my word choices.
    And I will do my best to stay away from this conversation.

  32. Okay, so when you put your claim like that, it does have a bit more merit, and I won’t argue the toss any longer. But here’s a pro tip for anyone who calls themselves an editor, or whose job has to do with communication in any form: When a whole lot of people claim that something you wrote was ambiguous, MAYBE THEY’RE RIGHT. And maybe you shouldn’t insist repeatedly that it isn’t. Just because you knew what you meant when you wrote it, that doesn’t mean you succeeded in making it obvious to anyone else.

    Thank you for making it clearer, but no thank you for insisting it always was clear.

  33. I appreciate your opinion, and I want to thank you for sharing it.

    However, I still don’t quite understand where the claim was *not* clear. It was in the original article, and we have been discussing it for the entire length of this page.

  34. “The Ambigram deck itself seems impossible at first, and in fact, it would have been impossible just a few years ago. No human being could have designed it alone, even if they possessed several lifetimes to work on the problem.”

    That’s the claim that Kevin responded to (that no human could have made a comparable Ambigram deck); it says nothing about word selection and word selection isn’t mentioned until several paragraphs later. It’s pretty clear that Kevin was responding only to the ability to make ambigrams out of an arbitrary word list, and that Editor agreed (at least before Kevin did it) that doing so was a reasonable response to the claim he’d made. It wasn’t until Kevin completed what was supposed to take lifetimes in 3 days that Editor decided that picking easily flipable words was the salient point, not designing a 7×7 set of ambigrams; if that was the important point he should have pointed that out before going along with Kevin’s challenge. Apparently the wording of the article confused even its author.

  35. Uhhh…. no….

    Yes, the original claim was that the deck would be practically impossible for a human. Yes, the elaboration was that it would have been impossible a few years ago (even for a computer). BOTH of these refer to the word selection.

    As I’ve stated at least 3 times now, the ambigram creation has been possible for 10 years. The word selection has not.

    Word selection was the claim from the very first paragraph. Yes, I did not go into detail in the first paragraph, but that’s the way writing works! You make the general claim up front, and then go into details further on.

    Kevin tried to disprove the claim that word selection matters by asking for a set of 8 random words that he would then create an ambigram grid from. He actually did a pretty good job (better than I expected) and I have already given him props several times here. However, he ended up actually further proving the point (even if it was just with the SAVOT design) that word selection actually IS important, and is the key to making an ambigram grid work (or any ambigram, for that matter).

    He proved the argument that with a set of random words, there will always be one (or more) designs that don’t match up well, and that finding a set of 7 words match up well actually pretty important.

    The claim has not changed, the argument has not changed, and the only reason the focus got off track at all is that some commentators on this page kept steering the conversation to focus on the design of the ambigrams instead of word selection.

  36. Interesting discussion here, so I’m going to add an experiment of my own. On my blog I posted an example of a full grid of ambigrams from a set of seven words chosen by a human—me, that is. I used a strategy mentioned earlier in the comments by the site editor:
    “I suppose you could “cheat” and choose 7 words that are composed of just the letters “i”, “t” and “l” (purely vertical), or “o”, “a” and “e” (purely circular), or that have the same basic repeating character patterns, …”
    I leave it to others to decide whether this method really is cheating and if I did a decent job with the ambigrams.

    Read the relevant blog post:
    or take a look at my solution:×7.png

  37. Hey Jutt,

    Congratulations! That is an awesome example.

    A theory is only valid until it has been disproven, and I do believe you have done it.

    I withdraw the theory that a human could not have designed a deck of cards consisting of a 7×7 grid of ambigrams worthy of game play. The legibility of your designs certainly makes it worthy of game play, and proof that a human is indeed up to the task of selecting the words to create a playable deck.

    It also is further proof that word selection is the critical component in the designing of such a deck, and you did a great job at selecting the initial words (even if it does use one of the techniques of “cheating” that I wrote about where a human could specifically select words that have the same basic repeating patterns to minimize the size of the task).

    Is it optimal? Who knows. Would it be “playable”? Yes, and that was the ultimate goal.

    I suppose the original claim should have gone into further detail. I didn’t actually expect so much “analysis” of the wording, as it was primarily an announcement about the game.

    I should have just stated that a human could not have done what the computer did, and left it at that.

    For example:

    • The computer had no restrictions on the letter set (the original set of cards considered all letters as being “in play”).
    • There was no restrictions on the length of the words (the original deck had words of any length “in play”, and ended up selecting words that had lengths of both 5 and 6 letters in the final set)
    • Finally, there were no restrictions that all ambigrams had to contain just one-to-one inversions (the original set selected many assymetric pairings)
    • .

    I still maintain that without the constraints that bound the challenge, the task does become “almost impossible” for a human, but I will admit that even with a restricted word set, restricted word length and restricted pairings (all one-to-one), a human CAN perform the word selection for a deck very worthy of game play, and ultimately, that was the goal.

    So, nice job! We’ll drop the claim so we can all move on with our lives. :-)

  38. The last post was in 2009, does it count if I start thinking about 7 words now so as to win this game or not? Thanx in advance. :)

(page contents are not an accurate historical record. Many pages were modified in 2012)
If you liked this page, you may also be interested in creating your own ambigrams or ambigram tattoos.