Reality Conditions

Thursday, March 23, 2006

Incompatibility: a Google game

There are a few games around, like Guess the Google, that are based on using the most popular search engine for recreational purposes. Following this tradition (and inspired by the weird searches that bring people here sometimes), I have invented a Google game, or perhaps better to call it a challenge. The name of the game is Incompatibility, and its challenge is to find two words that have individually a lot of hits in Google, but zero hits if searched for together. Incompatible words, so to say.

Some restrictions that I think are necessary to make the game interesting:

1- The words must be both in the same language (though proper names are allowed and considered to belong to all languages).

2- Both words should individually get at least 10,000 Google hits. This is to prevent tricks like using some obscure Russian name that appears only in a few pages in that language in conjunction with a very common English word like "the".

3- When I tried to start playing, I found that there are a few webpages, such as this one, that appearently try to list all words existing in all languages. I make a special rule to the effect that these pages don't count, so a search that gets only one or some of them is valid as "zero hits".

The best score I got after some time of trying was for the words: pharsalia supersymmetric, which gets 0 hits while the disjunction pharsalia OR supersymmetric gets 4,170,000. I guess that there is just no way of connecting a Latin epic poem and a symmetry relating bosons to fermions. By the way, their individual scores are respectively 963,000 and 3,120,000, which don't add to 4,170,000. I have no idea what Google is doing, but I will accept the results of the combined search with OR as official.

I have just realized that this game has a paradoxical self-defeating character: from now on there will be a Google hit for the "pharsalia supersymmetric" combination, this page! So perhaps a better way of thinking of the game, at least when its results are published on the Web, is not as discovering incompatible words but as creating, with an original act of mind, novel compatible words. I may have been the first person ever to think of Pharsalia and supersymmetry together, but from now on, the combination exists. Wouldn't you like to have that same honour for some pair of words? (Or perhaps you would, in a romantic fashion, prefer to keep some areas of the semantic space forever untouched by mankind?)

So, the challenge is set. Write your own incompatible pairs in the comments (even if they don't beat my score, I'm interested to see what you come up with). I declare the competition open for 5 days; on Tuesday evening, I will establish the winner as the first World Champion of Incompatibility!


  • Well, this somehow seems like cheating...

    IF by proper names you meant to include a person's first name, then I'm sure a winning entry should have the term oprah in it. As in:

    quasiconcave oprah --> leads to 3 sites defining economic (heh) terms [and you only find Oprah in the cached version, in a poll]

    quasiconcave OR oprah --> 26.5 M (though the first time I searched, it was 50 M -- apparently I found a bug)

    quasiconcave --> 27400

    I don't like this, bc one of the components adds many more hits than the other (guess which one!)


    quasiconcave psoriasis points to one list of words

    quasiconcave OR psoriasis brings 10 M hits.

    Again, I like yours better!

    By Anonymous Euge, at 12:32 AM, March 26, 2006  

  • I won't accept the first one as valid for the moment (until the cached version dissapears), as the two words were really on the same page which was not a list of all words. The idea was very good, though, and it is not cheating at all. It is true that a great "inbalance" in the number of hits of the two words makes the solution less pleasing, but it is perfectly in accord with the rules. Also there is no reason to exclude first names is surnames are allowed.

    The second one is valid, so at the moment Euge is holding the record with 18,100,000 hits for the search "quasiconcave OR psoriasis".

    By Blogger Alejandro, at 12:11 PM, March 26, 2006  

  • This challenge seems to be the opposite of what this paper is about, namely, finding words which have more joint entries than you'd expect.

    By Blogger david, at 1:25 PM, March 28, 2006  

  • Thanks for the pointer, David! It seems to me that the paper is being unfair to Google when they compare the results of their algorithm (which was searching over data in encyclopedias and datasets) with the results Google gets when searching on the whole, chaotic, messy WWW. Isn't the most natural explanation for their algorithm getting better results that the search space was "tidier"? Of course, if they claim only that they have shown the algorithm to be a well-designed and sensible one, not to be "better" than the one Google uses, then there is no quarrel.

    By the way, it is already Tuesday evening, so Euge has become officially the First World Champion of Incompatibility! The prize is a free anti-psoriasis treatment AND a textbook on quasiconcave functions.

    By Blogger Alejandro, at 10:13 PM, March 28, 2006  

  • Yes, I think you're right. Some of the queries they've chosen seen extremely well-suited for an encyclopaedia. I have the impression that Google Sets aims to find surprising sharply-defined sets, e.g., Jesus and Darwin as Cambridge colleges. Warrior and Soldier don't belong to such a set. What you want there are synonyms.

    When will we be able to enter 'Engineer' and 'Contractor' and be given a list of Indian international cricketers of the early 1960s (2/3 of the way down this page)?

    By Blogger david, at 1:15 PM, March 31, 2006  

  • Google Sets gives me an idea for your next competition - to use it to generate as bizarre a list as possible.

    Here's my entry: 'Rainbow' & 'Holiday' produce:

    Mascot, Masks, Historical, Wigs,
    Christian, Directories, Maps,
    Children, Imports, Audiophile,
    Household, Photographs

    By Blogger david, at 4:31 PM, March 31, 2006  

  • I have tried a little with your idea but couldn't come up with anything funny enough to quote... I guess that to make a real competition out of it, one would need first to establish a quantitative measure of "bizarreness"...

    By Blogger Alejandro, at 3:44 PM, April 02, 2006  

  • Babahoyo Weismannism have no hits go geather and Babahoyo has 12400 & Weismannism has 11,000 its not too hard, it took me....5 minouts to find it, and i am not the smartest 15 year old out there.

    By Anonymous Anonymous, at 11:45 PM, February 28, 2008  

  • By Anonymous Steven, at 7:55 AM, September 07, 2011  

Post a Comment

Links to this post:

Create a Link

<< Home