20 December 2009

Sensible nonsense

A fixture of the interactive web, these days, is the distorted text image which you have to type into a box (proving that you are not a robot) before continuing with some action. I have no doubt that the spammers and other assorted barbarians at the gate will eventually crack this; but not yet.

I am regularly impressed by the words offered on Blogger sites where I need to enter the magic password before my comments are registered. Not only are the words more easily readable than in many other instances, but they consistently manage to be pronounceable as well.

Both characteristics make transcription errors less likely, minimising the risk that I will stomp off in a sulk and not bother to comment at all. Thought has clearly gone into the system.

I'm not a natural or enthusiastic programmer. Perhaps the making of pronounceable nonsense words is trivial to those who are; but to me it's a small wonder.

I have written a spreadsheet which generates passwords for sites to which I must restrict access. It throws up twenty five groups of random alphabetic characters at a time, and I hit recalc until one of these groups is pronounceable. Since I set a single password for a period of time, and a usable group comes up fairly quickly, it's not an onerous process ... but it's not efficient, either.

Whatever algorithm Blogger uses is not just a simple matter of alternating consonants and vowels (though come to think of it, I might modify my spreadsheet to do just that). Over at JSBlog, for instance, I've just been asked to type in "madmast"; previous visits have asked for "svalploen", "omchak" and so on. I have seen "valvsnagorm" but I never get consonant groups like that "lvsn" at the beginning of a word. Clearly there is a fairly sophisticated rule base which specifies which letters can be used together in which circumstances.

As I type this, and think through the logic, it becomes clearer ... I still have no intention of trying to program such a mechanism, but I do begin to see how it would be done. Nevertheless, to see a different pronounceable but meaningless word pop up every time still feels like magic.


Ray Girvan said...

I very much like the reCAPTCHA system, that harnesses the process to the useful purpose of digitising old print texts.

Dr. C said...

I just posted a comment below and it was "hurandi" . I guess that's street talk for her and me