Well, you see, first off you need a microservice to distinguish between those 2 types of step, then you need separate microservices to handle storing those values. Then you need a GraphQL database in a multi zone Kubernetes cluster as a backend...
Ugh I feel dirty saying all that.
Just being able to exclude certain sites, and reduce the ranking of others, makes search so much better in Kagi.
I hope they eat Google's lunch.