Some solid ideas there, and let me say how I see those analogies play out in blockchain tech: Forking is still the current way to make radical changes (biological evolution). Smart contracts are the current analogues to the cultural evolution you talk about.
In theory, any improvement implemented via a smart contract can be implemented via a fork, and from a technical perspective (speed, memory efficiency) the forked implementation can be designed to run better.
But smart contracts do have on major advantage, and that's a political one: smart contracts because of the sand-boxed nature in which they execute (assuming no bugs in your smart contract processing code) enable new functionality to be added to a cryptocurrency network without requiring consensus to add the new functionality (only people who want to participate in the smart contract's new rules need to care about it, everyone else using the blockchain can ignore it). And here we see the analogy to cultural evolution when looking at smart contracts: changes can be made on a smart contract system without requiring change to the base platform (the DNA, so to speak). With a smart contract, you "teach" the existing blockchain how to do new things without making a low level change to how it operates.
Steem doesn't have smart contracts of this sort currently, and it would require a hard fork to add them, but it's certainly doable.
There's also another alternative to smart contracts as a means of evolution without requiring full consensus and that's by linking up chains via trustless crosschain transfer. Then users can transfer value between networks with differing rules, enabling them in many cases to obtain the benefits of multiple blockchains. This could be viewed as cultural evolution, or maybe it's more like "emergent behavior" (it's easy to stretch any analogy too far). But in any case, it's a way to enable new behaviors on a blockchain without changing it's DNA.
As a side point, note that "trustless transfer" technology is different from atomic crosschain swaps (although they commonly get conflated with each other), it's a different and more powerful technology, which to my knowledge has not yet been implemented for any pair of chains. But it's definitely theoretically possible and I expect to see someone do the hard work eventually.
But to achieve all the evolution of rules that I think we need to see in consensus technology (a term I'm coining now that is similar to what people call "blockchain technology", but only requires consensus mechanisms and no actual blockchain) I believe the performance advantages of hard-coded rules will be needed for some time to create technology that operates at sufficient speed to solve some important problems.
In a perhaps an ironic twist, my own primary area of interest is looking at ways to use consensus technology to speed up the rate of our human cultural evolution.