If strategy A is buying lottery tickets at $5 and strategy B is buying the lottery tickets at $10, A is twice as good off. It doesn't matter that the earnings are dominated by one single winning lottery ticket; strategy A is always twice as good as strategy B. So getting terms that are twice as good for a VC will double their expected earnings.
Sure, if you just consider the winning ticket, it doesn't matter whether you are paying $5 or $10 for it. The thing is that you don't know whats the winning ticket, and it does matter whether you are paying $5 or $10 for each.
I'm not getting it because the reasoning is not sound. Your reasoning is basically "the numbers are BIG! so a factor of 2x doesn't matter". If you can get your lottery tickets for half price, you can buy twice as many lottery tickets, and double your chances. Equivalently, getting 1% of Google or 5% of Google is a 5x difference, which is huge. That could easily mean the difference between a net profit or net loss for an investor over all his investments.
If you really believe that the terms don't matter, then I'm sure lots of YC startups are happy to take money from you at terms that are 5x better than what YC offers.
As it turns out, the numbers are BIG! so a factor of 2x doesn't matter. This is literally how it works. The only thing that matters in venture capital is being in the small number of companies (15 out of about 4,000 per year) that generate 97%+ of the returns. Those can pay off 1,000 to 1. Fiddling around on terms or bargain shopping or refusing to pay up for quality all reduce your odds of being in the winners, which kills your chances of winning as a VC.
Lottery tickets are a good example in this case. By definition, strategy A and B cannot both be winners (there is only one winning ticket). So the only thing that will set either apart is if one of the two strategies has the winning ticket. The price per ticket is only relevant if neither has the winner, in which case you're comparing who had bigger losses.
The reason I like the lottery example is that two venture capital firms don't have matching portfolios (if they did, then valuation would matter for relative performance). In reality, when you compare firm A vs firm B it's the performance of the startups that determines the winner (not the amount of equity owned).
Yes, in the end the VC with the winner is going to win, but that's beside the point even if there is a single winner in the world (which in reality is obviously not true, there are not dozens but hundreds of huge ROI winners). The point is that you don't know the winner beforehand. A VC who is getting 2*x% equity for $y is expected to perform twice as good as one that is getting x% for $y. The arguments that are being made here are incredibly vague. I'd love to see an argument based on solid logic why valuation doesn't matter much. I'm sure PG is right, but I'd like to understand why.
It's not the same equity. Quality is not distributed equally among the sample set of companies. The great startups (as judged retroactively via returns -- e.g. Facebook and Google) often (but not always) can get multiple competitive offers from top-tier venture firms, so having some abstract theory about how you're going to only pay low prices significantly damages your opportunity to invest in the companies that are going to generate all the returns.
Trying to maximize your share of individual deals causes you to lose the best deals. This is because a) the best deals are often expensive, and b) maximizing your share empirically causes people to consider you a dick, and people with reputations for being dicks don't get chosen by the best startups.
The point of this whole debate is this: VC return = equity * performance. Optimizing for the former is much less leveraged than the latter, since performance can vary by 10000x. If asking for better terms means you lose out on any deals then doing so is probably not in the firm's best interest.
Sure, if you just consider the winning ticket, it doesn't matter whether you are paying $5 or $10 for it. The thing is that you don't know whats the winning ticket, and it does matter whether you are paying $5 or $10 for each.