Whoops. You're correct. the Hexidecimal version of 2 is 2 <smile> I did mean those answers in jest as I realize most people would answer "two".

It's still a tough nut to come up with a large number of these kinds of questions that a parser couldn't pick apart and automatically answer. Ever play Zork? The ability to parse English text in context and take appropriate action is pretty amazing, and that was back in the 80s. It would seem that any cognitive puzzle sufficiently simple for an algorithm to generate (contextual or otherwise) would be simple for an algorithm to defeat. Likewise difficult problems end up blocking users as well as algorithms. We would appear to be between Scylla and Charybdis. The one area where this seems to differ is when we mix noise in. Noise that isn't generated by an algorithm is hard to separate. Sort of like trying to remove the sauce from the spaghetti. Luckily our own senses are very good at this and so the successful techniques leverage our ability to untangle intentional garbling of images and sound.

CB

Jane Lee wrote:
On Sat, Oct 4, 2008 at 6:20 PM, Jacob Schmude <[EMAIL PROTECTED]> wrote:

My point was simply that in the very complex languages it would be
difficult to have an answer to a question that could cover al the possible
answers. Sorry if I didn't make that clear.


But this is nitpicking and not going to come up most of the time. If it
does, it's a problem with the question and the question should be changed.
Here's my problem with the below quoted sample question and answers
mentioned previously:
The question:

If you have four apples and you give your friend two, how many apples do
you have left?


A contrived reply:

Of course the answer is "half", "less", "not enough", "What's an Apple?",
"two", "II", "2", "0x00010"..

(by the way, 0x00010 would be hex for 16, 0b10 binary for 2)

The problem with the reply: no English speaker in their right mind would
actually use most of those answers, otherwise they have a fundamental
problem in either comprehending the question or answering, in which case
they might as well be spammers. The assumption is that the question is in
English, and the answer therefore should be in English, and in English when
you are asked "how many apples do you have left if you had four and gave
away two" by occam's razor the two possible answers are the integer "2" and
a case-insensitive match of the word "two", which are trivial to implement
on the website's side (not thinking of the possibility of an integer and a
word as representations of a number is the fault of the programmer).

Yes, you ARE going to have some people submitting nonsense like the rest of
the "answers", but that is their problem, not yours, as long as your
questions are unambigious. Short of asking a question like "What is your
life story?", simple math (like the apples) and implied answers (like my
Apple-blog example before) and such similar questions should be more than
enough. If you have a specialized audience, you can customize your question
even more.

But how would that work for say a site like Paypal or another
banking-related one? Perfect for blogs an forums, though.


Paypal and banks and similar institutions have other methods. Bank accounts,
credit cards, showing up in person at the bank, phone numbers and more.
Blogs and forums don't have that kind of luxury.

cheers,
jane

Reply via email to