Totally agree with most of this blog, except I think sanding off chip numbers is a total waste of time and just makes you look like a complete amateur.
Unless you are using an EXTREMELY obscure chip, it won't take anyone with half a brain more than an hour or two to figure out what it is.
And as for potting, it really isn't worth the hassle unless it brings other benefits like environmental protection. Potting is a messy, time-consuming and expensive process unless you can afford the kit to automate the mix/dispense process. Don't think for a minute it will stop someone finding out what's inside, as a can of paintstripper will remove most compounds in a day or two.
And don't forget that X-Ray machines are as close as the nearest vet or dentist. I've done this once, and it was invaluable in determining where to start with the Dremel.
Microcontroller based products are by far the best way to add value while being relatively hard to copy. The great thing is that once you've spent the time writing software, and cramming in as many features as you can, it costs nothing to manufacture.
Niches are definitely the way to go. You will never make a penny selling to teh general public unless you get very lucky.
Here's a nice niche I filled a few years ago. This started off as a solution to a problem I had, Once I figured out a neat solution, I put out a few feelers on some forums to see if there was a demand, & ended up selling something like 3000 of them over a few years with zero advertising.
This was pretty much the perfect niche - not big enough for it to be worth anyone else copying, pretty much all of the value was in firmware, which costs nothing to manufacture. As with all these things there was also some luck involved, like MS launching Media Center.
Having a product out there can also lead into other products or consultancy work as you become known as an expert in hat particular niche. For example the above product led to me doing some custom hardware for companies testing set-top box software, and a lot of consultany in the area of infra-red comms.
I'm not yet convinced about the 'open-source everything' argument - It is very dependent on your target market as to whether it's a benefit, non-issue or potential liablity, but at the very least there is rarely anything to lose by having fully open APIs, interfaces etc. to allow other people to interface/expand/adapt your product. Someone may move your product into a market you never knew existed!
OK ready-made CC licenses are there, but you still have exactly the same problem as with patents when it comes to enforcing them - you might be able to persuade ebay to take down listings for infringers but any more than that you need to pay a
parasite lawyer, so I think that when deciding to go this route you pretty much have to assume people will disregard any licensing terms.