I was just looking at Taylor expansions today (guid uniqueness...birthday problem), so, this kind of reminds me of that. First you evolve, then you evolve the ability to evolve, then you evolve the ability to evolve the ability to evolve, etc. So maybe there's a formula.

Beyond that, what worked for trilobites largely works for horseshoe crabs.

Consider if something is getting larger by say 1% per thousand years, well over 1 million years that's simply unsustainable without becoming ~21,000x as large and thus a very different organism. Further conditions may change promoting periods of shrinking which means the average is going to trend to ever smaller changes per period of time.

AKA, fox sizes might drift up and down some over time, the average is going to stay near their current size unless say wolves died out. At which point they may quickly get larger and replace wolves, but they would not keep growing indefinitely and end up the size of great whales as they are now constrained by the 'wolf' niche. Further, foxes may end up spiting, you have foxes that look about the same, and foxes that look a lot like wolves.

Now if you chart sizes as say 10,11,12,11,10,9 then the transition between those stages is a 'fast' + or - 1, but the net result is only -1 over 6 steps or -1/6th.

The same thing is likely going on with less obvious features, but they are less intuitive.