Front end developer specialising in JavaScript and React. Experienced in all aspects of modern front end development. Passionate about making accessible, secure and performant software.
From one point of view, I think you answered the questions yourself. "There are usually better business gains elsewhere". You also implied that when performance does become a problem, it will be fixed. This seems like a reasonable approach to me from a business perspective.
About user behaviour (multiple tabs).. I think I would challenge this. In mobile I personally only use one tab at a time. On a PC, if I have 20 tabs open on YouTube I expect some slowdown. In other words, I'm not immediately convinced that this case is a problem. Maybe it is, maybe it's not.
One problem is that performance is not free. It costs additional code, potentially more complicated code, development time, and it can be finicky and error prone. E.g. with Redux if we don't use "reselect" properly, the result is worse than not using it at all (extra overhead from calling a function unnecessarily).
One way we could have better performance is if it is automatic (or close to automatic). This is one of Gatsby-js goals. "Performance by default". E.g. it is easy to have lazy-loaded images with Gatsby. Anything else is a business case, same as any feature in the software. Another way is if it's a single-time thing, such as setting up webpack.
How would we measure it? So far I've trusted statistics from Google and other large companies. E.g. the 3 second thing for loading. They have the ability to measure such things better than individual companies.
Interesting questions.
From one point of view, I think you answered the questions yourself. "There are usually better business gains elsewhere". You also implied that when performance does become a problem, it will be fixed. This seems like a reasonable approach to me from a business perspective.
About user behaviour (multiple tabs).. I think I would challenge this. In mobile I personally only use one tab at a time. On a PC, if I have 20 tabs open on YouTube I expect some slowdown. In other words, I'm not immediately convinced that this case is a problem. Maybe it is, maybe it's not.
One problem is that performance is not free. It costs additional code, potentially more complicated code, development time, and it can be finicky and error prone. E.g. with Redux if we don't use "reselect" properly, the result is worse than not using it at all (extra overhead from calling a function unnecessarily).
One way we could have better performance is if it is automatic (or close to automatic). This is one of Gatsby-js goals. "Performance by default". E.g. it is easy to have lazy-loaded images with Gatsby. Anything else is a business case, same as any feature in the software. Another way is if it's a single-time thing, such as setting up webpack.
How would we measure it? So far I've trusted statistics from Google and other large companies. E.g. the 3 second thing for loading. They have the ability to measure such things better than individual companies.
Thanks, these are very similar to my own thoughts.
The puritan in me keeps nagging tho.
It forces me to reconsider, my ideas at intervals.
Probably also why i practically answered some of the queries myself as you mention. :)