This is a known bug with a known fix. See Edge Tranches: Fixing an Incentive Misalignment. Basically, in the future I expect your “like cred” to become “use it or lose it”; i.e. you won’t benefit from being stingy with your likes. (People could still game it by only liking their own content, or posts that quote/reply to their own content, but this is a little easier to call out.)
The idea of having likes act as a mini boost sounds nice in theory, but it’s hard to see how to cleanly implement this. Boosting is rate-limited by the fact that it’s expensive to boost things; however likes are “free” so if likes were mini boosts, it would make an incentive to create tons of sock puppet accounts and like everything. An attempt to make the “boost value” of a like depend on the amount of cred that flows on the like edge would massively complicate the algorithm. Right now the edges and weights are fixed, and we use them to calculate cred. If we made the edge weight depend on the cred, now we would have an extra circular dependency in the algorithm, and it’s not obvious that it would be computationally tractable.
I would rather just double down on boosting, and make boosting things as natural and easy as liking in Discourse. If we wanted, we could even add a mechanism where every “like” auto-creates a small boost from the user. Each user could configure how much they want to boost per like they issue. So, for example, I could configure my Discourse profile so that every one of my likes corresponds to a 500 Grain boost. Someone with less Grain could set a smaller boost amount. (And if the person did not have enough Grain, the like would go through but the boost wouldn’t.)