[r-t] Candidate definition #10

[Mark Davies]

Don writes,

> Oops, sorry. I momentarily forgot I was replying to the fellow shouting at
> Pandora to keep her hands off the box. You don't want to allow five
> covers, for much this reason.

Indeed. The problem with the "blocks" idea is it doesn't allow some really
good stuff (interleaved multi-stage extent blocks) but it allows rubbish,
like

     row          pn
1   1234567
2   1243567  12567
3   2134657  7
4   1234657  34567
5   1326475  1
6   1326457  12345
repeat ad nauseam

being treated as a true peal of Triples, because it is composed of
true-and-complete 2-change blocks (rows 1,2 and 3,4 and 5,6 etc).

It's like the "Triples and Singles" problem, but a lot worse. By disallowing
stages <3 you could improve things, but you still suffer from the "too
strict" problems (1) and (2) and the "too lax" Triples-and-Singles problem.

Basically you've gone backwards with #10 Don - sorry - it's worse than our
preceding attempts.

MBD