So, I was reflecting a bit on the responses here, and two thoughts came to mind.
First of all, it looks like a "metaproblem" with GH1 is that mission payouts are linearly scaled to the PV of the opposition. Late game players who are embarrassing Joseph by rolling in the dough have pushed him to make the slope shallow and boost the cost of mecha. But that shallow slope means, for a starting non-combat-optimized player, mecha missions simply don't pay enough for the risk of mecha loss.
(They should pay much more, since the whole reason a player would accept a mecha mission is to turn a profit. If every penny of mecha mission pay goes into your Next Replacement Mecha fund and cannot be spared for, say, Robotics experiments, then you're really just doing slave labor.)
To fix this, we need to use a function of lower degree, such as a square root or logarithm. Then milk runs can pay adequately without getting crazy at the high end.
Second, with the point about the power player having all skills in the teens, that puts some new perspective on the players who feel compelled to buy every accessory combat skill (Spot Weakness, etc.). If Joseph intended it to be a tight decision whether to take Spot Weakness +6 or save the skill slot for something else, then it becomes very lopsided if Spot Weakness +15 is practical.
One other note about skills. Presently, GH1's skill cost formula follows an almost (1 1 2 3 4 5 8 instead of 1 1 2 3 5 8 ) Fibonacci pattern for the first 15 levels, but then goes quadratic. As Fibonacci is close to exponential, that's a decrease. If the skill costs were Fibonacci all the way through, absurd skill levels would be harder to attain. In fact, there would be an effective cap at about level 37, as the next level skill cost would then exceed what can be represented in 32 bits...