Based on the outcomes of T124959 let's make the necessary config change in the mobile stable channel.

• Make a statement about expected result of this change based on output of T124959. e.g. "I expect first paint to drop by 10% and fully loaded time by 20% for Barack Obama page"
• Make config change and SWAT it
• Perform an action=purge on the page and verify it is working
• Confirm whether your prediction was correct or wrong with reasoning.
• Note the impact (if any) on the global performance graphs and use this to determine whether the Barack Obama article is a fair representative of the work we are doing.
• Note impact on text cache for the week of the change which one would expect to trend downwards (less HTML servers)

The config change ideally should be coordinated after T125381 has been resolved (but not blocked by) as this will hopefully allow us to check the impact on a broader set of connection speeds.