DAE observation: Works better at higher speeds for scratched CDs. Technical reasons?

I have tested it with audio CDs and audio CD-Rs on several different vendor drives, and all had this effect:

When running DAE on a damaged disc at lower speeds, there is less noise, which is understandable. But there is a side effect: The drive is more likely to get stuck (hiccup sounds; unresponsive). On some drives, skipping is hearable on the extracted audio.

When running DAE on a damaged disc at higher speeds, although the extracted audio has more noise, the drive never gets stuck and gracefully DAE’s through the damage.

What is the technical reason for this behaviour?