I wonder if a compromise could be caching things only if they are widely used across public sites. A browser vendor could use telemetry or crawling to aggregate

I wonder if a compromise could be caching things only if they are widely used across public sites. A browser vendor could use telemetry or crawling to aggregate information about commonly used resources across the web. The browser could cache these resources, even proactively. It's certainly more complex than the shared cache, but it could achieve an end that is broadly similar. Then again, maybe the vendors' telemetry it telling them that first site load is not that common and that the shared cache doesn't move the needle that much. This wouldn't be surprising to find out.