This is very disappointing. Retiring the offsite part is very bad, it reduces drastically the expected returns for those investors that have trusted in your project for a long time, before Bustabit had the impressive track record that it currently has.
If you were using the offsite system as it was designed, you can easily just bring those funds onsite for free (with the dilution fee credits).
I think the running theme of your post is that early investors should have some sort of advantage over new investors. I get where you're coming from, and I say this as someone who literally invested in the bankroll on day #1 -- that I don't think that's fair. Earlier investors have
already been rewarded by with crazy returns. When they did invest there was probably a lot more risk, and there definitely was a hugely higher expected returns. I have a solid stake in the bankroll, and have quite literally withdrawn way more than I've ever put in. I guess I see it like bitcoin, it doesn't seem reasonable for early adopters to get some extra benefit/treatment for investing in the projects infancy. The reward is literally that it's worth so much more now.
Other than not letting people increase their offsite portion, the fair way to approach this situation would have been to gradually increase the dilution fee. This should limit the bankroll's growth without harming early investors. Another more extreme approach, but that you could consider, is simply not allowing new investments at all, but I think increasing the dilution fee is the fairest method by far.
Even if you did the more extreme version ( stop all new investments) the bankroll would continue to increase by the investor profit (minus divestments). Investor profit over the last 90 days has been 533.50 BTC, which is an insane amount. That's literally like $100k/day average. I assume Daniel is in the position where the bankroll has become so big, that the only thing that happens by it increasing is increasing his liability.
But if you think of it though, it's actually a pretty clever idea. I think it'll become the standard way of doing bankroll commissions in the future I think. Basically Daniel is running a sort of auction, where investors are basically bidding against each other to provide the bankroll bustabit needs. (Although for the record, I don't think a linear relationship is the most optimal way to model it, and I would've pegged it against USD or something a bit more elegant than a fixed number like the largest bet in the last 3 months)