Just out of interest and comparison. Below is a deliberate worst case on mine, pushed into the red and coming back down into yellow and green. Interestingly if you compare to the all-yellow graph, there is about twice as much swapping gone on compared to the graph peaking in red.
Obviously Memory Pressure doesn't equate swapping with automatic Red pressure, nor can it tell the future, only the user can tell if they are likely to run a Red-inducing workload regularly and therefore make an upgrade decision.
For an OS designed to fill memory, there has to be another way to advise on whether more RAM is needed, the pressure graph does the job pretty well for most users I think.
Honestly, at first, I was having a difficult time understanding what you were trying to say, so I wrote the following below as a way to explain how I see it. After finishing, I re-read what you had wrote, and I realized that a lot of what we're saying is the same thing, maybe just from slightly different perspectives...
There's a difference between
needing additional RAM, and
benefiting from additional RAM, just like there's a difference between literally "starving" and just being "hungry". When RAM pressure increases due to applications requesting more RAM than is currently available, there is a dip in performance as memory is compressed and/or paged out to swap. How noticeable that is depends on the computer hardware and duration of memory pressure until the memory manager can free up RAM. In most situations, for average users, minor moments of memory pressure aren't going to be very noticeable, particularly on an SSD-equipped system.
The Memory pressure gauge indicates duration and intensity of memory pressure at the current moment in time (and shows the last few minutes of recorded memory pressure). So as you suggest, swap by itself is not an indicator of memory pressure at any given moment... there are scenarios where a small amount of swap is needed here and there, that accumulates over an extended period of time, but never having much perceptible impact on performance... or there could be a very large spike in memory pressure that results in a large amount of swap being used all at once, and system performance stalling, before memory management compensates and reduces the memory pressure back to green.
So there's really a case to be made both ways: The Memory pressure gauge may not indicate "past" lags in performance at any given moment, and swap and memory compression may not indicate the
severity of lags in performance. Unless observing in real time, it's hard to know.
I think we can all agree that too many unknowledgeable users either equate performance issues with lack of RAM (when it's often some other issue) or look at their memory usage and misinterpret what they're looking at. When there are spikes in memory pressure, it's really up to each user to determine whether lags in system performance are noticeable to them, and whether they're really worth addressing (sometimes with a little help of more knowledgeable users).
