This is precisely why we proposed a system with adjustable parameters to strike the right balance. We also suggested starting with downtime slashing events only, as a test case. Another idea we support is to have the penalty only partially covered, ensuring that this remains entirely voluntary for validators. We still believe delegators should face slashing for validator misbehavior; however, offering an optional incentive mechanism tied to the validator bond presents a potentially valuable solution that warrants exploration.
We fully acknowledge this valid concern, and we agree it must be handled with great care. Fine-tuning the parameters is crucial. As referenced in @arlai-mk’s calculation, a 0.01% slashing event with a 100% fraction (which we do not necessarily recommend) would cost a validator like Crosnest nearly 13 ATOMs. This counters the argument that decentralization would be hampered. However, you’re absolutely right—careful adjustments are essential, particularly if this feature extends to more costly slashing events. We believe the idea of linking validator bonds is worth further consideration and in-depth discussion.