How to exactly is having filters making it more difficult expensive to run nodes leading to centralization?

Taking Bitcoin’s 80‑byte OP_RETURN cap off is like removing the mesh from a window—spam won’t vanish; it just blows straight in.

The line “e‑mail is 90 % owned by five companies” actually counts reading apps, not mailbox hosts. Apple Mail, for instance, only fetches messages that can live on any server. Hosting is more concentrated than it used to be. However, it’s no monopoly, and the plumbing remains highly federated: the Internet is still dotted with independent MX servers and live SMTP nodes. Deliverability is concentrated; the transport network is not—so e‑mail’s path doesn’t prove Bitcoin must centralise if a policy knob disappears.

Hashcash failed because SMTP lacks Bitcoin’s two brakes: an unforgeable fee token and a hard block‑space ceiling. Spammers could mint proof‑of‑work “stamps” on hijacked PCs for almost nothing, while every Bitcoin transaction must burn real satoshis and bid for finite space. Remove the OP_RETURN cap and, when fees are low, jumbo payloads will bloat the chain’s permanent history; when costs rise, attackers will bury data in off‑curve pubkeys, inflating both history and the UTXO set unless those keys are made invalid.

Until consensus rejects off‑curve keys and the fee schedule fully prices storage (e.g., via a higher vbyte multiplier for data outputs), the 80‑byte limit is the least bad option. Lifting it now would swap a miner‑relay concern for a permanent chain and UTXO bloat that every node would carry forever.