I. CAP Theorem and Blockchains
II. Privacy vs. Transparency
III. Incentives (this post)
Decentralized systems must deal with spam. Email spam used to be a widespread problem (all profitable channels eventually become saturated) until (1) email reputation systems and (2) centralization.
Blockchains provide a novel approach to spam through consensus algorithms like proof-of-work and proof-of-stake to generate transaction fees.
Blockchains add transactions fees in exchange for avoiding the cost of reputation-building and centralization.
Today, it's not difficult to run your own email server (I ran my own in college), but it is difficult to ensure your mail is reliably delivered.
The cost of centralization doesn't show up as a monetary cost borne directly by the consumer– it manifests itself through the downsides of monopoly power – possibly fewer incentives to be efficient and lower consumer surplus.
Without high enough transaction fees, there's not a viable incentive for third parties to validate transactions (or capture Miner Extractable Value) on-chain. Without high enough transaction fees, spam would be difficult to combat – there's no centralized service to rate limit or block bad actors. The chains that have approached zero transaction fees still have some fees. Otherwise, spam would be rampant.
But I believe that when things are too cheap to meter, they should become free. And that the upsides of centralization have the potential to be much better than decentralization for both consumers and network designers.