Editorial Policy

Last updated: August 1, 2026

This site analyses public companies in industries I work in professionally. That creates obligations most blogs don't have. This page sets out the rules I hold myself to, so you can judge the work by a standard I've published in advance rather than one I invent afterwards.

Sourcing

Everything factual on this site traces back to material anyone can access. In practice:

• Company filings, earnings calls, investor day presentations, and press releases
• Granted patents and published patent applications
• Peer-reviewed research and conference proceedings
• Manufacturer datasheets and published specifications
• Reporting by established trade and financial press, cited and linked

If I can't point to a public source for a factual claim, I don't publish the claim. Where a number is derived rather than reported — a calculation, an estimate, a range — I show the inputs and the arithmetic so you can disagree with it.

Scope

I don't cover companies I have a working relationship with, and I don't use non-public information — not as a source, and not as background. Where that rules out a topic, I skip the topic.

How deep I go on process technology

This site stops at the question of why something is hard and what that costs. It does not walk through how to do it. You won't find process recipes, tool parameters, integration schemes, or anything that reads as an implementation guide, even where such material has been published somewhere.

That's partly a legal boundary and partly an editorial one. The readers here are investors, not process engineers. What matters to them is why only a few companies can do a thing, how durable that advantage is, and what it does to margins — not the recipe.

Fact, estimate, and opinion

These are different things and I mark them differently. A sourced fact carries a citation. An estimate says it is one and shows its working. An opinion is written in the first person and flagged as judgment — usually with language like "I think" or "my read is". Where my professional experience informs a judgment but can't be cited, I say that explicitly rather than dressing it up as established fact.

What this site is not

It is not investment advice, and it is not a stock-picking service. I don't publish price targets, ratings, or buy and sell recommendations, and nothing here is tailored to your situation. See the full disclaimer for the complete statement.

Conflicts of interest

Where I hold a position in a company I write about, I disclose it in the article itself. I don't write about a company for the purpose of moving its price, and I don't trade around my own publication schedule.

This site carries advertising. Advertisers have no input into what I cover or what I conclude, no advance access to articles, and no ability to have anything removed. I don't accept sponsored posts, paid placements, or payment for coverage. If that ever changes, this page changes first and the disclosure appears on the article.

Corrections

I will get things wrong. When I do, I fix it in the open rather than quietly editing the page.

• Factual errors are corrected in place, with a dated correction note at the foot of the article describing what was wrong.
• Substantive changes to an argument are marked as updates, with the original reasoning left visible.
• Typos and formatting are fixed silently — no note.

I don't delete published articles to hide a bad call. If an argument doesn't survive contact with reality, the more useful thing is to leave it up and write about why it failed.

How articles are produced

I'll be direct about this, because you should be able to weigh it: drafts on this site are AI-assisted. I use AI tooling to survey public sources, assemble the evidence, and produce a first draft against a standing brief — the topic scope, the sourcing rules, and the constraints on this page.

Nothing reaches this site on that basis alone. Before anything is published:

• Every factual claim is checked against the primary document — the filing, the patent, the paper, the datasheet. Not against a model's summary of it, and not against secondary reporting.
• Every source is opened and read. A citation that doesn't support the sentence attached to it gets the sentence removed, not the citation swapped.
• The engineering judgment is mine. What a constraint actually implies, whether a demo proves what it appears to prove, which number is the one that matters — that is the part I'm here for, and it doesn't come from a model.
• Automated checks run first against the rules on this page — minimum sourcing, citation density, and the exclusions above — and a draft that fails them doesn't reach review.

AI output is never a source. If a claim can't be traced to a public document I have actually opened, it doesn't get published, regardless of how the draft arrived.

I'm telling you this because the alternative is letting you assume otherwise. A site arguing that you should check what's underneath confident-sounding claims doesn't get to be an exception to its own standard.

Independence

Views expressed here are mine alone. They do not represent my employer or any organisation I'm affiliated with, and nothing on this site is published on anyone's behalf.

Questions and challenges

If you think something here is wrong, tell me — corrections from readers who know a subsystem better than I do are the most valuable feedback this site gets. See the contact page.

— Tony