I’m writing this post quickly for a more permanent record of developing thoughts. This style and many of the ideas were spurred by a few blog posts Wilson Cusack wrote last year, which I recently read and reference below. Highly recommend reading them too!
In Wilson’s Fees for the Good of the DAO, he argues that protocol DAOs should seek to only take fees where it makes the protocol stronger. I've started referring to these kinds of fees as "productive fees."
One example Wilson cited is a hypothetical: what if Uniswap charged a fee to flash LPs that hop in-and-out to provide liquidity for a single transaction. The motivation: 1) earn the protocol revenue 2) make the protocol stronger by advantaging long-term LPs who make Uniswap a good place to trade. I'm not sure of the merit of this specific idea, but taken at face value, I think it illustrates the idea of "productive fees" well.
Wilson's second post, The Case for Value Capture at the Periphery, makes the case for value capture at the application layer:
Clearly DAOs are exciting and probably needed for many things, but I think that many parties taking fees at the periphery, and the core team working to make this highly viable, is under discussed. It strikes me that the dynamic of many productive and profitable teams wanting to see the core keep working might be more generative than many people trying to vote their way to a share of one giant pot.
Products create a lot of value for their end users, and as Wilson argues, it is healthy for protocols to have an ecosystem of sustainable applications building on top and driving protocol usage.
The most practical way for this to happen is for those products to be able to sustain themselves through a free market enterprise. This seems plain as day, but it is something I’ve actually diminished in previous writings suggesting that DAOs seek to commoditize their complements (the products building on top) by subsidizing their development.
Commoditizing a complementary product in order to bootstrap a protocol is a good playbook at the outset. The most succesful DeFi protocols like Uniswap and Compound gave away their products for free, to help kick off the network effects of the underlying protocol. But this model can't last without protocol subsidies to fund the product. I've come around to the view that mid to long term, this approach is suboptimal because it is too bureaucratic.
Instead, an optimal equilibrium may be 1) protocols sustain themselves through "productive fees" that make the the protocol better, faster, stronger such that products want to build on top, and 2) product teams charge "extractive fees" that capture the value they create for end-users.
Tying this thread to another, I think there's an argument that the more products built on top of a protocol, the more likely protocol DAOs will be able to charge "productive fees." Lets use Uniswap as an example, which has a rich ecosystem of apps on top driving demand to the protocol, many of which do extract their own fee (e.g. Metamask,) In aggregate, that demand helps keep Uniswap's monopoly on supply in tact, as there are strong network effects to LPing where the demand is routed. Today, Uniswap has a simple fee switch, which may actually diminish the supply-side network effect. But if Uniswap were to update its tokeneconomic model as described in the hypothetical model above, it could help reinforce a strong moat on supply and keep the flywheel between products and protocol spinning.
You can contrast this with protocols that have a a single or more power-lawed product ecosystem. An example is OpenSea (product) vs. Seaport (protocol.) In this example, and others like it, value accrues to the company building the product.
I wrote a thread last week, comparing this market dynamic to that of NFT collectors who value creator royalties. The tl;dr: even if you are a product team who values decentralization, if you control the bulk of value flow, the market will make it hard to justify that value flowing to a DAO instead of the company creating it.
I was discussing all this with Peter Watts from Reservoir, and he raised a great point, that perhaps one additional consideration for protocol value is the degree to which the protocol can successfully monopolize supply.
As I wrote above, if demand is distributed at the product level, but supply is monopolized due to network effects (e.g. Uniswap) you may be in the sweetspot for protocol value. However, if supply is able to easily multitenant AND there is fierce competition for demand at the product level, you may end up in a race to the bottom in both product and protocol, as is increasingly the case in NFT marketplaces. The former seems more likely with peer-to-contract markets, and the latter more likely in peer-to-peer markets.
Of course, reality isn't so theoretical, and if you aggregate lots of users, or lots of developers, there's likely a way to step into the flow. But its an interesting thought.