Still, this is an interesting issue, as optical kerning values in ID are connected with type size you are using in the text sample. If you are going under 72 pts, kerning values will going smaller and smaller. Also seems, ID will correct supposed kerning values endlessly, if you will check kerned pairs there again. Sure, “delete small kerning pairs” is a must run after “stealing”.
Yes, optical kerning is dependent on font size in InDesign. A good first run would include scaling it to a size where the n-n and n-o combinations have a kern value of (close to) zero, and then run the script.
After that I round kerning values by 5, delete small kerning pairs (<10), and compress kerning.
Are you rounding them manually? For deleting small kerning pairs is a script.
Can you try the latest cutting edge version (2.6.2-1234). I hope I fixed the Apple Script problems.
There is also a script for rounding. It’s called Adjust Kerning in Master and has a few options. Rounding is one of them.