One problem I see is that most gadgets could just be replaced by a modern mobile phone. For example the fitness thingy from the article: what does it do that an iPhone app couldn't do with a nicer interface? Same for internet radios - an iPod touch is just 200$.

"Today with a guy or two, a good idea and about $1.5 million you can get a contract manufacturer in Asia to do your gadget."

So probably not very easy with only a few thousand in investment, but I guess somewhat reasonable.

It's possible to mock up a hardware design in software (just look at console emulators). That gives one a cheap way to demonstrate ideas that can be used to attract investors. Plus, the software version is a useful platform for testing new ideas and identifying hardware bugs.

If your hardware idea is radically different from any existing device (rather than using software to make the hardware unique), you can still use physical models to simulate actual mechanical inputs (like accelerometers). Otherwise, you can just find an existing hardware platform and hack it to run your software.