Richard McDaniel

This analysis misses how courts actually treat infrastructure vs publishers. The "notification = mens rea = liability" chain breaks down because: 1. Transmission = Storage in practice Every router, ISP, CDN, email server, Tor relay, and cache stores data. If storage meant criminal possession, the entire internet would be a felony distribution network. Courts have never accepted this. 2. Safe harbor exists for "can't delete it" systems Tor relays can't inspect onion cells ISPs can't filter encrypted packets E2E messengers can't delete from devices CDNs auto-mirror content until notified Bitcoin nodes fit this pattern perfectly. Courts treat these as infrastructure, not publishers. 3. Courts avoid absurd results Judges consistently refuse interpretations that: Criminalize ubiquitous technical operations Destroy entire industries Impose impossible compliance standards If deletion is technically impossible, courts treat that as a technical fact (like routers unable to inspect encrypted traffic), not criminal intent. 4. The publisher assumption doesn't hold His argument requires nodes be "in control" of content. But courts almost never assign publisher status to: Automatic replication systems Decentralized protocols Intermediaries lacking editorial control 5. Public interest dominates Bitcoin serves journalists, dissidents, NGOs, institutions, and governments globally. Courts ask: does node liability produce more good or harm? We don't have criminal Tor operators, TCP/IP router operators, or email server admins for the same reason we won't have criminal node operators. The law isn't a suicide pact. It adapts to technological reality.

At one point, someone hired me to develop a proof-of-concept trojan for educational and defensive research purposes. When it became clear they were discussing things that sounded legally questionable, I cut ties and refused further involvement. Not long after, I received an unexpected message from the FBI saying that someone using both my handle and my real name was linked to cybercrime activity, and asking whether I knew someone known as “Iceman.” Later, it emerged that Max Butler, who went by Iceman, had adopted the alias Aphex among others.

Out of billions of images from 2002, perhaps 10⁴–10⁵ can be cryptographically proven today. Most are lost to bit rot, dead hard drives, or format obsolescence. Aphex.jpg sits in the 99.9999th percentile of verifiable digital continuity, authenticated by multiple independent malware databases with timestamps and checksums. But preservation alone isn't the story. This single JPEG has FIVE independent layers of historical significance: Layer 1: Rosetta Stone-tier digital artifact One of ~100 images from 2002 with forensic-grade provenance. Survived because antivirus companies became unwitting archivists. The adversaries preserved what the creator didn't keep. Layer 2: Early malware history Part of W32.Aphex@mm worm (CNN, 2002). Created by a figure in the Windows rootkit scene, featured in cybersecurity textbooks, credited with discovering Sygate and Tor vulnerabilities. Layer 3: Bitcoin genesis Aphex was User ID 369 on Bitcointalk (hundreds from Satoshi), received forwarded email from Satoshi via Eugen Leitl, founded first legally registered Bitcoin exchange (Bitcoin2Cash), created first physical Bitcoin bearer instruments (predating Casascius). Layer 4: Cultural crossover Aphex Twin (Richard D. James) released a track titled "W32.Aphex@mm" on Analord 09 (2005), directly referencing the worm by name. One of the most influential electronic musicians of all time immortalized this malware in his discography. Layer 5: Blockchain permanence Inscribed on Bitcoin via Ordinals. The same network Aphex helped pioneer now permanently hosts the artifact. Narrative comes full circle: early Bitcoin adopter → Bitcoin-native digital artifact. Each layer alone creates museum-worthy value. Together they're multiplicative, not additive. Try to name another artifact, physical or digital, that's simultaneously: Provably from a specific historical moment Created by a tech pioneer (Bitcoin exchanges) Referenced by a major artist (Aphex Twin) Preserved through adversarial infrastructure Inscribed on the technology its creator helped establish There isn't one. This isn't a PFP. This isn't a memecoin. This is the digital equivalent of a verified fragment from Pompeii, except it exists in computer security history, Bitcoin history, digital art history, AND electronic music history. The cave painter didn't think archaeologists would study their handprint. I didn't preserve the JPEG, VirusTotal did. Survival itself became the rarity. Now as long as Bitcoin exists, so does this piece of internet mythology. theoriginalpermissionlessjpe…

Okay, yea you’re cool. But do you have your own Aphex Twin song?

AI doesn't destroy art. It reveals authorship. When technical skill becomes instantly replicable, what’s left exposed is intention. The brushwork, the rendering, the polish, all the traditional veils of mastery dissolve, and what remains is the why. That’s uncomfortable, because many artists unconsciously lean on craft as both armor and identity. If anyone can produce a flawless image, then what differentiates you isn’t execution but vision, the coherence of your taste, your worldview, your story. Art stops being “look how hard this was” and becomes “look how true this feels.” That re-centers the artist’s inner landscape as the real subject. You can’t hide behind process anymore. If you have nothing to say, the machine will amplify that emptiness.

15 years ago today, I joined a tiny forum of Bitcoin enthusiasts. User ID 369, just a few hundred registrations away from Satoshi Nakamoto himself.

The current frontier of agentic coding tools appears to be the summary-loss problem. To handle long contexts, these tools rewrite conversations into summaries that serve as the new foundation for continued reasoning. But summarization often compresses away details that seem minor or repetitive, like how to run tests, why certain commands are preferred, or what each file does. As a result, the agent “forgets” information it had “learned” just steps earlier. One potential solution is self-organizing, dynamically generated memory clusters formed from the content and patterns of the interaction itself. The agent infers the structure of your working environment, then clusters information around latent concepts: tools, workflows, gotchas, preferences, and so on. Over time, it can update, merge, split, or prune these clusters, just as you would with your own evolving mental model of a project. This approach addresses the summary-loss problem by preserving semantic structure rather than flattening rich procedural memory into a generic blob. The agent improves the more it works with you.

Not that it matters but BTC already outperformed NVDA if I’d held. One generational trade into the next.

Just every single corporate AI converging on the same outcome without coordination. Probably nothing.