Interesting idea, for sure!
Here’s what seems to me like a major issue: Time/maturity and fund redeeming. This fund naturally invests in early stage companies. It can take a very long time till you see any return (or even partial liquidity). Further, if this is an ETF (which would be the natural instrument to get the benefits you’ve mentioned), then as investors sell their public units (for whatever reason), sometimes (if there are no immediate buyers for them in the market), the fund needs to redeem their units and return the funds. Where would those funds come from? The fund owns stock in 1000 early-stage startups — and there is very little liquidity for these assets. You could sell some of them — but those are likely to be the ones you actually want to keep. Plus, managing this process would require human judgment.
Alternatively, you can construct it as a closed-end fund, but then you’d lose some of the important advantages you’re trying to get here.
One other question I’d ask is about this assumption: “Due to the 2/20 fee structure, limited partners of funds can only realize above market returns if the fund invests in one or several extremely high-performing startups (100x returns on a single startup).”
I’m no VC, but from the math I’ve seen, this need for home runs stems from the high failure rate of startups, not from the management fees. And if that is true, then the suggested solution does not actually solve the core issue, since the batting average of such a fund will likely not be higher than that of other VCs/syndicates. In fact, due to factors you and Chris Treadaway both mention, the batting average is likely to be lower, exacerbating the issue.
One last thing about the concept in general. IIRC, most VCs are not outperforming the stock market. The ones that are outperforming — what do you think makes the difference? Possible answers I can think of are — access to deal flow, deal selection & due diligence process, and value-add. Yet these are the exact factors such a seed fund would be weakest at, not so?