Sunset the Web Worker Offloading plugin

[westonruter]

Feature Description

The Web Worker Offloading plugin was an experiment to see if we could leverage Partytown to improve INP. However, I have yet to see any reports that it improves anything. By contrast, I've only seen poor reviews and problems in the support forum. I suggest we remove Web Worker Offloading from being listed among the Performance features to install.

There are currently 20K+ active installations of the plugin right now. To gather feedback from users to find out if they are finding it improves INP in any way, we can perhaps show a banner on the top of the Performance screen advising users that we're considering turning down the experiment. If they find it is useful, we could ask them to submit feedback to the support forum.

We can add a notification dot next to the Performance menu item if they have the Performance Lab plugin active. If they don't have the Performance Lab plugin, we'll have to figure out an alternative way to notify users of our plans to sunset. Maybe we'll have to show the banner at above the plugin list table.

[adamsilverstein]

I'm fine with deprecating experiments that don't work out - thats why they are experiments! we probably need a documented process for that anyway.

Questions:

• Is there any data we can look at to confirm benefits?
• Is there any harm when the plugin is disabled on a site it is currently used on? (Guessing no)
• I like the idea of a notification - can we add that for two releases, then just disable the plugin entirely.

[westonruter]

• Is there any data we can look at to confirm benefits?

I don't have any data. Just anecdotal. Given the very uncertain benefit of this, I think we can at least start with just removing the plugin from being listed as a feature. I don't think we should encourage it to be installed if we can't vouch for it.

We can then collect feedback from existing users who have it installed whether it is shown to be improving anything. There are 20,000+ active installs so that's a good sample to get feedback from. I don't think we're well equipped anymore to analyze those active installs to see if they have better INP than sites that don't have it installed.

Another point of concern regarding Partytown is the most recent release was back in July, it's still not a 1.x release, and activity has all but ceased:

• Is there any harm when the plugin is disabled on a site it is currently used on? (Guessing no)

No, there shouldn't be.

• I like the idea of a notification - can we add that for two releases, then just disable the plugin entirely.

I'm not sure we should forcibly disable the plugin, but we could maybe add some inline style or inline admin notice to the plugin in the list table to alert that it is deprecated and should be uninstalled. That is, assuming we don't find that actually INP is improved with the plugin.