I'm surprised to not hear an engineer's memory as an important correlator to skill. I find that working memory especially (or the lack thereof) really can hinder my performance when dealing with a complex problem. Sometimes I think it is THE defining factor!
If it's been 6 weeks since I've worked on/with x and now I need to work with x again, what's important is that I have a process to reproduce x or build on x. I'd argue this is a better practice than trying to commit everything to memory because human memory is pretty corruptible. I also find this helps with complex problems because you don't have to deal with the devil upfront, you just have to be able to do some high-level fuzzing.
That being said, I think it's important to commit mistakes to memory because those are embarrassing to repeat.