[QUOTE=davieddy;279233]Of course most factors are found at or below 71 bits, but the time needed to do this is the same as that for raising the limit from 71 to 72.[/QUOTE]The odds of finding a factor per bit level goes down each bit level (~1/71 vs ~1/72) while the effort doubles. Here are the stats for the factors found so far in the range quoted above.
From 29 bits to 38 = 2922 factors.
From 39 bits to 48 = 2401
From 49 bits to 58 = 1470
From 59 bits to 68 = 829
From 69 bits to 78 = 270
From 79 bits onward = 6

It seems that a 'broad' search eliminates far more exponents per GHz-day than a 'deep' search. The crossover point of more TF vs LL saved is a different issue. But the little bit of effort to take a 100M digit to 68 bits (less than 2 hours on this machine), is the most valuable. 73 to 74 is ~60 hours. That is hardly "practically worthless".

All davieddy is saying is that as far as total effort goes, we are but a very small fraction of the way to setting up 'fully factored' LL tests. He's not disputing that this very small amount of work has eliminated by far the most exponents.

That is to say we've found 95% of the factors that will be found, but that only 10% of total TF effort to the breakeven level has been done.

[QUOTE=Uncwilly;279239]But the little bit of effort to take a 100M digit to 68 bits (less than 2 hours on this machine), is the most valuable. 73 to 74 is ~60 hours. That is hardly "practically worthless".[/QUOTE]
2 hours for 29 to 68 bits. "Little bit of effort"
4096 hours for 29 to 79. "Large bit of effort"
4094 hours for 69 to 79. "That's better. Shame about fewer factors
found, but it would have taken me another 2 hours: That "little effort"
to take it to 68 was really worthwhile - [B]NOT[/B]".

Are you still determined to misinterpret/contradict me?

I'm talking about LL tests (GIMPS) here,
not the
"Great Internet Low-Hanging-Fruit-Factor Search"