Currently I am building a new machine, call it 2.0, that does something slightly different than machine 1.0 but closely related. So I am drawing on the knowledge that I gained from building machine 1.0, duplicating as much as I can. However I also get the chance to fix all the design decisions that I completely boggled up the first time around.
There is one piece on my machine 1.0 that has a tendency to shatter. It gives me excellent performance, but that's just poor design on my part. Shattering is not an acceptable mode of operation, unless perhaps I am building missiles. And since building a potato gun is as close as I have ever come to making missiles, designing something to shatter is not on my list of appropriately funded activities.
So I have been redesigning this piece of my machine for version 2.0. But I have found myself stuck. I have spent a week working on this, and it has come down to several choices.
1) Buy the thing perfectly suited to what I want, which is an experimental cutting-edge product just barely on the market (of course, I do research, nothing I want is going to be COMMON or EASY to FIND...). Cost: about $2000, while my current shattering piece solution is only $200.
2) Make the thing I want myself, which I think is doable but tricky and unknown. It potentially involves some very fancy knowledge from well outside my specialty. And my fear is that "tricky" may turn into "uh-oh this is not doing what I want, oh crud it's going to take me five weeks to debug this, crimeny this is way more complicated than I expected!" Then we have a cost of materials: $200, time: 5 weeks, and student frustration: 2 handfuls of pulled hair.
3) Buy something commercially available that doesn't shatter, but doesn't quite give me the performance I want. This gives cost: $200, research contribution to the field: 0.
To add a final twist, other people are waiting impatiently for machine 2.0 to use themselves, and they don't need the cutting-edge performance I am really driving for. They just need something to WORK.
So perhaps, come to think of it, I should go with option #3, while I work on option #2?
Comments, internet hordes?