My general unlock path is Mk. II harvesters (both types) at the start, then at about 4-7 planets in (depending on map layout), Mk. II econ stations (replacing some of my Mk. I econ stations I placed in planets that are moderately safe with Mk. IIs). Pretty much never go for the Mk. III harvesters or the Mk. III econs, the Mk. II econ and harvesters provide all that I reasonably need.
Grab a calculator, or an excel spreadsheet and just check how much resources you are losing from picking ecoII instead of HarvIII.
You will see that even if at 7 planets you build all 6 of your eco stations, HarvIII's would still give about 200 m+e/s more, and they would scale, and not take any space. All that at the cost of additional 1k knowledge.
tl;dr MKIIs suck. Dont keep those, always go for a MKIII. The K cost is high, but the advantage over MKII is even higher.
Then that is part of the problem.
If the Mk. IIs of both paths have a lackluster resource to knowledge rate, then that is strong incentive to complete (aka, go to Mk. III) one path. With knowledge pressure being what it is, often times this means you can't complete the other path. And since one both paths' Mk. IIs have a low rate of return per knowledge, this means it isn't worth stopping midway through the other path.
So this leads to the question of which path to take. If you are planning on only taking 12 planets or so, then the econ station path will give you better returns. But the harvester path will scale beyond that, not have an opportunity cost, plus it will give you better rate during that critical 0-3 planet time period. Thus many players will go the harvester route.
Seems like either Mk. IIIs need to be nerfed some, and/or Mk. IIs need buffed some, such that going midway and saving a bit of K, while maybe not hyper optimal, won't put as far behind in terms of resource per knowledge as it would now.
EDIT: Generally speaking, AI War has had a tradition of having Mk. III and above have, while not terrible, rather poor rate of return for their utility per knowledge, for these very sorts of reasons (where the Mk. III+ and up may have a great effect, the knowledge cost goes up even faster still) From what I understand, the idea behind this is that, generally, if you want Mk. III+, that means you are willing to sacrifice variety for quality for a specific strategy (this goes with the whole trying to promote variety design goal). I think this should apply for harvesters and econ stations (and the other stations) as well.
EDIT2: Oh, and for my last game, having an extra 1k knowledge was worth losing 200 m+c/s. I'm not a fast player, so I don't lose money very fast, so that little bit more of income wouldn't be worth it to me. Plus, in my last game, I was very pressured for knowledge. Every 500 knowledge counted.