The procudure with the old table format would have been that the mapping
was from the 64 bits of keystream to any of the rightmost coordinates
on the register state plane and the lookup result would have to
be backclocked 100 times. When a grey cloud would have been hit by that,
the lookup result would have been useless. And a hit in a green cloud
would have been as useful as before. That means each hit not in a grey
cloud would have yielded 13 states on average but you would hit a grey
cloud 13 times more often than a coloured cloud.
To rule out all the grey clouds, the 100 extra clockings have to be done
during table generation (if you are in a grey cloud and clock forward
100 times you cannot be in a grey cloud anymore of course, at least
one of your paths can be clocked back 100 times).
Also if you clock the whole picture one more step forward, it would look
quite a bit different, because with each forward clocking you open up
3 more possibilities to clock back (the majority rule allows 4 different
clocking patterns). Each clocking forward would shift the current picture
as is one column to the left (changing the colors only gradually) and
open up 3 new planes with a basically quite different appearance.

If you ignore the extra planes spawning, you would see green clouds
in the original plane fading towards a single red path with no bulge in
the middle, red clouds with a bulge in the middle would have the same
destiny and you would see new green and red and orange clouds form in the
grey clouds. Red clouds would turn green as their bulge hits the left side.
Also, those single red paths would constantly flow out
of the picture into different y coordinates in the keyspace, but they never
end.
_______________________________________________
A51 mailing list
[email protected]
http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51

Reply via email to