What is in a filter list

An ad blocker is, at heart, a list. It does not know what advertising is — it knows which addresses deliver it. Those addresses live in filter lists, and how good a blocker is depends in large part on which ones it ships with and which of those are switched on after installation.

There are two kinds of rules, and they do very different things.

A network rule says: this request does not go out. The browser compares every address a page wants to load against the list and drops it before it leaves the device. Nothing is loaded, nothing is rendered, nothing costs time. That is why pages with a blocker are usually finished sooner than without one.

A cosmetic rule says: this element is not to be seen. It applies where advertising has no address of its own — a sponsored post in the middle of a feed, for instance, arrives over the same connection as the rest of the page. Cosmetics do not prevent the loading, they tidy up afterwards. The difference matters: a network rule saves you data, a cosmetic rule only saves you the sight.

The large lists are built in the open. Volunteers add addresses, others check them, and when a rule breaks a site it is taken back out. This is unspectacular, and it is the reason the whole thing works at all: no single company could keep up with the pace at which ad networks change their addresses.

Regionally it gets interesting. Advertising is not the same everywhere — what is served in Germany comes from different servers than what is loaded in Brazil or Japan. That is why there are regional lists alongside the general ones, and a blocker that does not switch on the right one blocks systematically less in that country.

And then there is a limit that is new. Since Chrome moved to Manifest V3, an extension may only have a limited number of rule sets active at once. Anyone shipping more lists than the platform allows has to choose — and what is not chosen does not block, however good it might be. More lists are therefore not automatically better. What counts is which ones are switched on.

That is also why we name no rule counts on our pages. A number sounds like performance and says nothing: twenty thousand rules with half of them switched off block less than five thousand that are running. What counts is the result — and that can be measured instead of claimed.