The root disposition - substance over hype

The root disposition that anchors every theme: name the substrate, refuse the rebrand, evaluate at the layer below the label.

Reduce any hyped category to its substrate, then evaluate.

This is a procedure, not a posture. When a new label arrives: name what it actually does at the layer below the brand. If the substrate is genuinely new, the excitement is earned. If the substrate is borrowed technology with a better-sounding name, refuse the frame. Same move, eight years, four distinct cycles.

The procedure in the corpus

2018, blockchain: "Blockchain is a database innovation first which can be used to improve a lot of processes." Written while running V2 Games, a blockchain startup - which is what makes it harder to dismiss as outside criticism. The disposition is not external contrarianism. It is internal honesty under financial exposure.

2018, ICOs: "Somewhere in between blockchain and ICO the product is getting lost. ICO is a fund raising tool which helps visionary teams build product for the future." The cleanest archetype: separate the funding mechanism from the product claim. These are not the same thing.

2024, Devin: "I would catch on to the hype once Devin replaces its founders to build Devin 2.0." A substrate-test in conditional form: withhold belief until the claim proves itself at the level it is actually being made.

2024, GenAI: "Almost everyone believes GenAI is overhyped. Yet there is something magically different from other techs that were hyped and never lived up (I am looking at you Blockchain)." The substrate-test passes: GenAI radically reduces the cost of cognition. That is genuinely new. The hype is earned. The 2018 callback is intentional - same test, different verdict.

2025, product customization: Every customization setting is a design decision deferred to the user. The hype-word is "flexibility." The substrate is implementation cost - six months, a partner, hundreds of training documents.

Why this is the root

Every other theme on this site applies a version of this test to a specific domain. Agent-first applies it to AI architecture - agents as substrate, not chatbots with UX skins. Enterprise AI reality applies it to deployment - demo vs. production as the real test. PM taste applies it at the PRD level - anti-customization is the same disposition executed in a product decision. Breadth as differentiation applies it to career strategy - Human-GPT as the failure mode of breadth without a real substrate underneath.

This page does not apply the filter. It names the filter.

Where to go from here

Three themes where this disposition is load-bearing:

• Agent-first thesis: the 2025 substrate-evaluation that declared horizontal AI dead and named agents as the new layer - the clearest case of the procedure running at scale.
• Enterprise AI reality: nine takeaways from production deployments, converted from "AI is exciting" to "AI in production is brutal" - substance-over-hype as field data.
• PM taste: anti-customization as the disposition executed at the PRD level, where the hype-word is "flexibility" and the substrate is cost.