[Corinthian]

I agree with quantum_mechani on this that the problem was not exponential investments, so much as it was the no brainer situation. Because lets face it: In a strategy game, EVERYTHING you spend resources on is an investment. You invest in troops to get provinces, you get provinces to get gold, you get gold to get troops.
The gem gens was too cheap and gave too good a return on the investment. It became far superior too more conventional early game strategies, like spamming low level summonings.

In the early game there is not a lot of spells that will put you in a better position, than hoarding gems, for casting higher level spells later, will do.

Also, clamming was not obvious to your neighbors, unlike a high province or gem income count. Witch would be the result of other kinds of investments. Therefore it prevented people from gauging your threat level and gang up on you before it was to late.

All of these things could be fixed by reducing the profit of the investment. In vanilla, a clam of pearl will pay for itself in 14 turns and could be made with hammers after researching const 2.

My solution is to make the gemgens worse investments. Lets change the cost of the clams from 15W 5N to 20W 5N and also change the construction requirements to const 6. Also if you are going to return the hammers, give it the "No forge discount" tag. This would change the rate of return from 14 to 25 turns. And by the time you reach const 6 there will be other high level spells to compete for the gems.

It will still pay for itself eventually, but it is no longer a no brain investment and will probably only be done by well entrenched turtlers, like underwater nations. Or by people that wanted const 6 for other reasons.