See it for yourself

Nothing on this page is sent anywhere. It runs in your browser, and it will keep working if you turn your connection off.

Please read this first. This is not the real infrastructure. It is a reduced toy version with none of the real mechanics in it.

It is here so you can watch the behaviour with your own words rather than take our word for it. The real demonstrations are shown after consent, and by request.

Two pieces of writing, and how alike they are

Type anything into both boxes. The bar shows how much the two have in common. Then try the buttons underneath and watch what moves and what does not.

0% 

 

The same, twiceIdentical writing always gives an identical answer. Not usually. Always, on any machine, today and next year.
A small changeChange a word and almost everything stays. That is how a near copy is recognised as a near copy rather than as something new.
Something unrelatedTwo unconnected pieces of writing land far apart, so a thing is never mistaken for a thing it has nothing to do with.

What this is standing in for

In the real system, that recognition is what lets a store keep one copy of what it already holds and only the difference for what is nearly the same. It is also why the same question always gets the same answer, and why an answer can be checked rather than trusted.

On the toy above, the numbers are the toy's. The measured figures for the real system are in the deck, and every one of them names where it was taken.

What it refuses to do

It will not guessGiven nothing, it says nothing. An empty box gets an honest refusal rather than a confident number.
It does not phone homeTurn off your connection and try again. Your words never leave this page.
It is not the productThis is a toy standing in for a behaviour. It cannot do the work, and it is not meant to.

Ask for the real demonstrations