Let's discuss some positive feedback loops:
- Funding decisions take into account numbers of publications and the prestige of journals where publications appeared. But in the authors-pay publication model, you need money to publish, and prestigious journals are more expensive. There are even researchers who buy authorships, although such cases are considered fraudulent, and are presumably still rare.
- Increasingly, conference organizers fund neither participants, nor even invited speakers, so the possibility for them to accept the invitation depends on already having money. For example, this is the case at the most prestigious mathematical conference -- the International Congress of Mathematicians. But attending conferences is helpful for obtaining funding: conferences allow to meet colleagues who may contribute to funding decisions, and invited talks at prestigious conferences appear on CVs and therefore funding applications.
- Grant applications can be so complicated that professional help is required. So research institutions are hiring consulting firms and/or dedicated staff for helping their researchers apply to grants from the European Research Council.
The existence of positive feedback loops is only one indication that the existing research funding system is wasteful and inefficient. There is a recent fashion for laureates of the Nobel prize to explain that their prize-winning research could not have happened in this system.