Yeh, my idea was only "fair by audit" and not "provably fair" (so it only works if investors trust that the auditor is not colluding with the site owner.) I actually made a diagram of that too some time ago:
The obvious disadvantage is performance. It is a bit like MP apps (they also make a "Make bet HTTP request" for each bet to MP) - they do work reasonable fast AFAIK, but it is still a disadvantage.
Most ETH dapps will probably still have low limits if truly provably fair for investors too. Some might use ETH blockhashes but that limits the bets to 0.05 BTC (due to block withholding.) Some use BTC bridge to use BTC blockhashes but that limits bets to 12.5 BTC (unless long calculation like PevPot) and I am not sure how those ETH sites handle BTC block orphans. Something like vDice is not really provably fair (even for players) as it uses third-parties for the RNG (random.org with some audit), probably "good enough" for 100 ETH bets, but for 100 BTC bets that seems too risky.
A P2P game like Contingency does solve the problem, but it is relevant to say that every bet is only connected to 1 investor/individual. So this still limits the max profit of each bet (way) more than "traditional crowd-funded sites" (since only 1 investor will need to cover the whole bet profit) and will result into some investors having a profit and some others a loss (more randomly than normal = more variance not very attractive for investors.)
Overall, definitely no real good way IMO
The obvious disadvantage is performance. It is a bit like MP apps (they also make a "Make bet HTTP request" for each bet to MP) - they do work reasonable fast AFAIK, but it is still a disadvantage.
Most ETH dapps will probably still have low limits if truly provably fair for investors too. Some might use ETH blockhashes but that limits the bets to 0.05 BTC (due to block withholding.) Some use BTC bridge to use BTC blockhashes but that limits bets to 12.5 BTC (unless long calculation like PevPot) and I am not sure how those ETH sites handle BTC block orphans. Something like vDice is not really provably fair (even for players) as it uses third-parties for the RNG (random.org with some audit), probably "good enough" for 100 ETH bets, but for 100 BTC bets that seems too risky.
A P2P game like Contingency does solve the problem, but it is relevant to say that every bet is only connected to 1 investor/individual. So this still limits the max profit of each bet (way) more than "traditional crowd-funded sites" (since only 1 investor will need to cover the whole bet profit) and will result into some investors having a profit and some others a loss (more randomly than normal = more variance not very attractive for investors.)
Overall, definitely no real good way IMO