Oddbean new post about | logout
 Oh no…I was just able to reproduce it when using the Microsoft Precision Mouse when connected via Bluetooth. But it takes around 60-70 windows visible on screen to make it happen (as opposed to ~25 when connected via USB). 
 With 70+ windows visible, it even reproduces with the ancient Logitech MX300 USB mouse. 
 @a6ff0b2b Maybe it's related to the mouse polling rate and the Microsoft mouse connected over USB has a higher polling rate than the Logitech or when connected via Bluetooth? The higher the polling rate the fewer open windows it takes to reproduce the issue. 
 @bcccefa6 I can get it to happen with the ancient Logitech MX300 which has to have a very low polling rate. 
 @a6ff0b2b In my work with xScope and Notchmeister, I have noticed that global event monitors have a lot of CPU overhead (they sniff a lot of event traffic). Maybe one of them has snuck into NSWindow or NSView?

https://developer.apple.com/documentation/appkit/nsevent/1535472-addglobalmonitorforeventsmatchin?language=objc 
 @a6ff0b2b I can already hear @f0727292 and @3d174fcc preparing an entire segment dedicated to ‘Issues only John would find’. 
 @a6ff0b2b
 🕳️

🕳️

🕳️

🕳️

🕳️ 
 @a6ff0b2b Based on anecdotes I heard on podcasts I’m somewhat surprised you initially tested with fewer than 70 windows. 
 @a6ff0b2b This might not be closed as "Could not reproduce" but rather as "Way too complicated to properly reproduce". 
 @7d1cd9eb It's trivially easy to reproduce! You just need an affected mouse, of which there seem to be many. 
 @a6ff0b2b have you factored in yet possible interferences from the neighbor’s garage door opener? 😅 
 @a6ff0b2b brilliant analysis - nevertheless, you use your computer wrong ;)