How a state actually slows a technology

When people imagine a government slowing a technology, they imagine a ban. Bans are rare, loud and easy to route around. The instruments that actually change what gets built are quieter, and three of them do most of the work.

Export controls

A rule that something may not leave the country without permission. The permission is the point: a licence regime is not a ban, it is a tap, and it can be opened or closed without passing a new law.

What makes them powerful is where they are placed. A control on a finished product is easy to design around. A control on one purification step, or on the machine that performs it, catches everything downstream — including products made in other countries by other companies, if the rule reaches that far. That is why the list of controlled items so often reads like industrial trivia.

The signal to watch is scope creep at the tool layer. A control on a material is a price event. A control on the equipment that makes the material is a capacity event, and those last years.

Trade defence

Tariffs, anti-dumping duties, safeguards. These protect domestic producers from imports, and they can help or hurt the thing you care about depending on whether the domestic producer actually exists at the grade you need.

This is where the most instructive failures happen. A measure written to support domestic polysilicon drove away both customers of the only sizeable domestic plant, which is now at risk of closing. The intent was to loosen a bottleneck; the effect so far has been to tighten it. That is not an argument against trade measures. It is an argument for checking which grade of a product a measure covers, because "polysilicon" is two different products and only one of them goes into chips.

Permission to build

The least discussed and often the most binding. A grid connection queue, an environmental permit, a reactor licence. Nothing is forbidden; it simply takes years, and there is nothing to buy to make it faster.

The interesting thing about this category is that it is the one where reform demonstrably works. When a regulator changes how a queue is processed — first ready, first served, with deadlines — the queue can move. That is why this site records process reform as an instrument alongside export controls, and why one of its published predictions is about whether a specific reform will work.

How to read the Policy section

Every instrument carries what it does, to which bottleneck, on what date, and one sentence from its own source. A status that is not simply "in force" has to explain itself: adopted but not yet in force, under investigation, suspended.

And where a named organisation publicly asked for a rule, it is recorded — but only where that organisation says so in its own document. Inferring who lobbied for what from who benefits is how a research project turns into a conspiracy board, so this site does not do it. Exactly one instrument currently has a named backer. The rest say so plainly.