# Yet another framing on Attention MEV

## Thesis

MEV as a universal pattern wherever control is delegated.

MEV is fundamentally about control delegation. When users trade sovereignty for convenience, they create extractable value that third parties can capture. This pattern exists across both crypto and social media platforms.

### In Crypto

When you broadcast a transaction, you delegate ordering/execution power to:

- Searchers who scan for arbitrage opportunities
- Block builders who arrange transactions
- Validators who finalize the ordering

Your transaction becomes an "agent with an open API" - it can be simulated and executed under various conditions, allowing others to extract value from its execution context.

### In Social Media

When you consume content, you delegate your brainfood curation to:

- Platform algorithms that rank your feed
- Influencers who shape discourse
- Ad brokers who insert sponsored content
- Bot farms that manipulate engagement

Your brain has an "open API" - every piece of content is an executable that auto-installs and runs on your neural pathways, shaping your thoughts and behaviors.

## The Control-Value Exchange

In both systems:

- **Users trade sovereignty for convenience**
- **Delegated control creates extractable value**
- **Financialization allows this control to become an asset**, entering free market competition that theoretically reduces extraction

| **Crypto** | **Attention** |
| --- | --- |
| Blockspace | Information Diet (timeline) |
| assets | vibe/culture/theme |
| addresses (private keys) | accounts |
| smart contracts (non-EOA addresses) | flashapps (encumbered accounts) |
| transactions (agent that can be simulated and executed) | user attention (brain with open api) |
| arbitrage | x-posting |
| Spam on-chain searcher | Bot-farmers that do theme page accounts, “natural traffic” |
| Off-chain searcher paying builders | Paying MCNs, influencers (you are sending orderflow to their partial block) |
| Off-chain searcher paying validators (mevboost PBS) | “Paid traffic” on platforms (platform ads) |
| Bundles/partial-blocks | Accounts/lists/follows (each account is a mini-timeline) |
| private mempools, RPCs that guarantee execution quality (e.g., flashbots protect) | Ad blockers, bluesky starter packs, timeline tuners, timeline filters, good aggregation accounts to follow (delegated control for some kind of execution/curation quality guarantee) |
| blockbuilders | MCNs (they find the efficient packing of a subspace of the entire attention space and they don't fundamentally control the distribution but they act as a good part in packaging it) |
| Self-building bundles | Users themselves creating a playlist/follow list |
| fair ordering - removing MEV via sequencing rules & pushing it to bundler/orderflow level | Platform ads algorithm sucks and everybody just pay influencers/MCNs or spam their own accounts for natural traffic, pushing ads upward the supplychain to the account level (which then turns into dark influencer deal games / spammy theme accounts) |
| flashbots | if TikTok didn’t have platform ads and outsourced to third party team |
| PBS | if TikTok integrated MCNs it owns with the algorithm it owns |
| statistical MEV | predictive algorithms shaping user behavior + simulation based on private data |
| buying orderflow (proprietary access to execution of transactions) | buying accounts (proprietary access to distribution of attention) |
| vertical integration of block builders and orderflow providers for offchain execution/rfq | move from public domain to private domain attention flow (brand spin up accounts to “own” distribution rather than “renting” it every time by paying platform for ads) |
| vertically integrated builder settles rfq on their own chain | private domain attention funnel to the attention operator’s own app |
| vertically integrated builder settles rfq on an L2 | private domain attention funnel to wechat groups (wechat is the L2) |
| L2 sequencer debate | if new social platform has no ads and is asking people to provide opinion to how their algorithm should work |
| an L1 with a good mev mechanism packaged as a product | TikTok (a platform with a good curation mechanism - the algorithm- packaged as a product) |
| balance between bundlers/orderflow originators or apps like cowswap and validator/blockbuiler | balance between account and algorithm (platform controlled) |
| **At worst, you lose money** | **At worst, you lose yourself** |

## The Philosophical Implications

### Consciousness as MEV

*Your 'self' is just whatever entity successfully extracted the most value from your attention inputs.*

- We're not autonomous individuals but "consensus states of competing executables"
- Our identity is shaped by whichever attention extractors most successfully programmed our neural patterns
- Unlike blockchain transactions, **there's no rollback for consciousness state transitions**

### The Fundamental Tension

Both systems struggle between:

- **Efficiency** (timeline algorithms/sequencers provide better UX)
- **Sovereignty** (decentralized control preserves user autonomy)

Users can restrict access control to prevent extraction, but this often results in worse execution quality - less relevant ads, less engaging content, higher transaction costs.

## The VERY IMPORTANT Market Structure Differences

### Crypto MEV Evolution

- **Started decentralized**: Hundreds of searchers competing
- **Platform (Ethereum) took little**: Just base fees
- **MEV democratized first**: Individual searchers could compete
- **Then consolidated**: Flashbots organized the chaos
- **Platform capture came last**: Validators started taking MEV share via PBS

### Attention MEV Current State

- **Started centralized**: Platforms monopolized from day 1
- **Platform takes everything**: Meta ~$135B, Google ~$238B in ads
- **Scraps for others**: Influencer marketing ~$21B (less than 10% of platform ads)
- **Already consolidated**: You literally can't compete with the algorithm

**It's as if Ethereum launched with built-in Flashbots that kept 90% of MEV and ran as a for-profit company!**

## Actual Market Sizes (Reality Check)

### Platform MEV (The Monopolists)

- **Meta ads**: $135B (2023)
- **Google ads**: $238B (2023)
- **TikTok ads**: ~$18B (2023)
- **Total platform extraction**: ~$400B+

### Non-Platform MEV (The Scraps)

- **Influencer marketing**: $21B (generous estimate)
- **MCNs**: Maybe $5B globally
- **Bot farms**: Unknown but probably <$5B
- **Affiliate marketing**: $17B (but mostly flows to Amazon)

**The ratio is 95% platform, 5% everyone else**