Stream is right, the difference is in this subproject the sieve was prepared specifically to give a higher chance of twin and Sophie Germain.

Note about other subprojects:

Twins in other projects: In projects where we already search both the +1 and the -1 form (for same series), like 321 and all four (current) subprojects on PRPNet, a twin prime would be found automatically, you would just have to notice it.

Sophie Germains in other projects: In base 2 projects where we use the -1 form, like The Riesel Problem, these would be found automatically. This is because for each k we do, we do all n. So if k*2^n - 1 and k*2^{n+1} - 1 were both primes, we would find both; you would just have to notice it.

In base 2 projects where we use the +1 form, we could find a pair of nearly doubled primes, k*2^n + 1 and k*2^{n+1} + 1. These are similar to Sophie Germains and would go to Caldwell's Cunningham Chains (2nd kind) page as chains of length two.

So while such finds could theoretically be found in (some) of the other subprojects, it is unlikely because the sieving there was not designed to boost the chance of that.

Addition: Note that for some projects or some k values, divisibility by 3 alone is enough to exclude the possibility of twins and/or Sophie Germains.

/JeppeSN

Crun-chi, but what data are you scanning? From what subproject? /JeppeSN

Let b be a fixed base not divisible by 3. Let k be a multiplier.

Twin: If for one exponent n, both k*b^n + 1 and k*b^n - 1 are primes, then k must be divisible by 3.

Sophie Germain: If there are two consecutive exponents that give primes k*b^n - 1 and k*b^{n+1} - 1, then k must be divisible by 3.

Cunningham chain 2nd kind: If there are two consecutive exponents that give primes k*b^n + 1 and k*b^{n+1} + 1, then k must be divisible by 3.

So in most of our subprojects, we must have k divisible by 3, in order for a discovered prime to be able to be twin (to the side we can test deterministically), be Sophie Germain/safe, or be in a Cunningham chain of the 2nd kind.

Our GFN primes may be in a Cunningham chain of the 2nd kind when b is divisible by 3.

/JeppeSN

Let b be a fixed base not divisible by 3. Let k be a multiplier.

Twin: If for one exponent n, both k*b^n + 1 and k*b^n - 1 are primes, then k must be divisible by 3.

Sophie Germain: If there are two consecutive exponents that give primes k*b^n - 1 and k*b^{n+1} - 1, then k must be divisible by 3.

Cunningham chain 2nd kind: If there are two consecutive exponents that give primes k*b^n + 1 and k*b^{n+1} + 1, then k must be divisible by 3.

So in most of our subprojects, we must have k divisible by 3, in order for a discovered prime to be able to be twin (to the side we can test deterministically), be Sophie Germain/safe, or be in a Cunningham chain of the 2nd kind.