Did you check the equation? There are two scenarios here:
-
Most/all delegation is self-stake, example: a validator with 5% VP launches 5 validators of 1% VP each. Well, he doesn’t have to do it in the first place, because according to the formula 5% (VPi) - 5% (the VPsb^2/VPi), so the VP tax would be 0
-
Most delegation external, not self-stake: this other 5% VP validator could try to launch 5 new validators, but good luck attracting this 5% VP to the new validators out of the active set. If he manages to do this, according to the formula VPsb would be almost 0 so can be ignored, now let’s assume median is 0.17% and honestly with this 5 new validators the median is unlikely to change much so. 5% - 0.17%= 4.83%, and in the case of 5 new validators, 5*(1%-0.17%) → 5*0.83=4.15%, not very different
Edit:
I checked also what you mentioned about dynamic commission @Sephiroth, what this does is forcing a high minimum commission on the largest validators. The idea here with VP tax is similar, since this tax would be like a higher commission for the largest validators with low self stake. The difference is that in the case of the VP tax, this tax is then distributed equally for all validators to allow them to run many consumer chains. In the case of the dynamic commission idea it is a partial solution, there is no tax collected or distributed equally, just a hope that this higher minimun fee would lead to redelegations and increase decentralization. The VP tax gives results immediately after being implemented, at the least by distributing this tax to all validators and fostering the growth of the AEZ, and in the best case on top of this reducing centralization over time, thus increasing the security in the Cosmos Hub and hence its value