Amadeus,
D6DS is definitely involved twice- I got that from Cordain who says he got
it from Simopoulos, and that is the current state of knowledge. As you say
D4DS is also involved but is less important. It is critical to get long
chain and short chain omega 3's in the diet. The elongation process is
hopelessly inefficient and there is a lot of inter-individual variation
(just read a few abstracts on Pubmed about that). I think the solution is to
maintains the paleolithic intake of LA LNA AA EPA and DHA, and that is easy
to do by using olive oil, eating lost of greens,  eating fish, grass fed
meat and organ meats- try to eat liver or kidneys once or twice weekly-
brain is even better but I can't stomach it (yet). Will send you the EFA
pathway separately.

Here's something I wrote to someone else:
Delta 6 desaturase (D6DS) is in a bad position as it has 4 potential
substrates- LA/ LNA/ C22:4n-6/ and C24:5n-3. This occurs because D6DS acts
twice in the pathway. This means that the it is easy to block it up by
having an excess of ANY of these substrates.

The modern diet has been glutted with LA. Excess LA or even LNA will clog
the enzyme and stop it from working on the C22 and C24 substrates. So large
doses of flaxseed oil aren't going to necessarily end up giving us large
amounts of DHA after bioconversion.

So bad enough that LA and LNA both compete for D6DS but in excess either of
them will make it very hard for it to act on longer substrates and make DHA
etc.

But of course a Paleo diet doesn't have any of these problems. I use the
Paleo diet as the skeleton for my study as it seems that everything fits
into it neatly.

Udo Erasmus' book Fats That Heal Fats That Kill is very good, he is a little
bit too angry, and the book lacks the practical pragmatism and real-life
usefulness of The Omega Diet. Erasmus is more like a reference book you use
when you want to find out about something e.g. grape seed oil.

Regards
Ben