Thedutchguy87 wrote:As I'm adamant (for the moment) to stay on paper,
Cool. Not many of us around, I guess.
I'm going to try the b/b plot you suggested.
I wouldn't recommend that. A full b/b plot (the way I understand it) is way too cluttered to be useful.
It should be better than my 9 mini grid approach.
I don't think so. The single-digit grids will help you much more than a b/b plot ever could, even though (or because) they have less information. They're invaluable for finding single-digit patterns and chains, even if you have a good replacement for b/b plots, so don't give up on them. They're just not enough for finding 3D AICs, so you need other tools as well. For me the single-digit mini grid is usually the first level of escalation if I feel I need more help after completing my main grid mark-up (which has the power of b/b plots but with less clutter).
The GEM question rephrased would be: Is there a more efficient way of marking up candidates/strong links?
I'm not sure if I understand the question. How do you define "efficient"? If you want to solve a non-basic puzzle as quickly as possible without computer aids, I'd say GEM is about as efficient as it gets (if you know how to use it efficiently). I just don't see it as a general way of marking up candidates and strong links, so your question seems moot to me. GEM is a coloring technique, and very efficient and effective at that.
I have created a comprehensive way of marking up candidates and strong links, but I don't see it competing with GEM at all. My system provides visual aids for following links by eyeballing but it's still the player's responsibility to find the useful chains. GEM, on the other hand, finds eliminations/contradictions directly, from which you can then deduce the chains/nets responsible for them. Mine is a direct replacement of b/b plots but GEM isn't -- mainly because the results of the latter depend on the chosen seeds, i.e. you may need several GEM mark-ups to get the full picture.
Personally I think GEM is an awesome technique when really needed. However, I think it's over-powered for most normal sudokus because it takes away the joy of finding useful chains on your own. Then again, it's a very nice trick to have in your toolbox for harder puzzles when your own eyeballing no longer produces results (or if you want to solve faster, or if you're just learning to find chains). I have created a quick way to use GEM with pencil and paper, so it certainly does not require software helpers. I personally use GEM as a last resort method, but I'm not saying it's the only way to use it.
(and additionally should I bother with weak links in any way?).
Generally, no. You need to be aware of them, especially with XY-Chains and other ALS chains, but marking them all up would produce too much clutter (like the full b/b plot). Concentrate on identifying strong links and learn to imagine the weak links between them. In some helper grids with just subsets of the candidates (e.g. bivalues, single digits) it may be useful to mark the weak links as well.
. Just for clarity it might help to give a little background into my solving: I use AIC's allowing grouped and ALS nodes. For especially hard puzzles and whenever I happen upon them I also use Denis Berthier's Z and T extensions. 
