You will get a lot of responses I am sure, but there is a glaring omission from that video. How are you going to monetise your efforts? It just says the house costs money to build, but it conveniently doesn't say how that open source architect pays the bills.
Howard, that assumes that the only way to make money on is to keep the everything secret and sell the licenses to use it. That's certainly one way to do it, but not the only one. Not since a long time.
Ask companies like RedHat, SuSE or even Github how they are making money from open source. They aren't selling the code but services around it. Theoretically anyone can grab the same code and become a competitor, in practice it needs certain specialized knowledge and skills - which is what these companies are happy to sell you. Not everyone wants to become an expert on Linux servers, for example - so you pay a specialist to set it up for you. And whether or not the "code" is closed or open source does not really matter for the client at this moment - they are paying for the service. The code being open is an advantage from the support point of view, though - they are not locked in to a particular vendor and it also makes debugging of problems easier.
This "give the code away and charge for services" model is getting to be a more and more prevalent, because the "code" is becoming a commodity (except in some domains requiring specialized knowledge/skills). There are only so many ways one can build a database or a webserver, for example, so you aren't going to make a killing selling those (unless you are Oracle - but that's more because of industrial inertia and corruption than because of the qualities of that software). And once that happens, there is no advantage to keep such code secret. In fact, it will only put you at a disadvantage because you are not going to get the bug fixes and security problem alerts from third parties which your competitors may be getting.
Now, I am not saying that you must open source your products. That's is up to everyone's decision. It requires a different mental model where you must stop thinking about the possibility of someone else "ripping off" your code because it is out there free to grab but need to have a clear idea how do you outcompete them instead, so that the customer spends the money with you and not on a cheap clone.
You must know exactly where is the added value which you are bringing. If all the value of your product is in that code, then you probably shouldn't open source it, because you will starve. On the other hand, if you have a sensible ecosystem around it and the code is not really *the thing* that is making the most money, it makes sense to open it up. It could bring you new business you wouldn't have had if the tools were closed and people weren't able to use them beyond what you have envisioned.
Just look at the embedded development tools - thanks to gcc/gdb being open, it is a defacto standard compiler/debugger for embedded programming. You would be hard pressed to find tooling that doesn't support it. That, in turn, has made life simpler for vendors of debugging pods, analysers and what not, because they know that if they support this combo, they will be able to sell their product no problem. The same for chip vendors - they don't need to build and maintain their own compilers and IDEs anymore (which was never a profitable business), just use gcc and voila, problem solved. Win-win for everyone, except for the sellers of the proprietary toolchains ...
Now when it comes to open sourcing hardware designs - that's a bit different story. With software it is simpler, with hw you may have those external constraints (like NDAs) which may prevent it. Fair enough. Concerning adding support for something like Linux in your product - just ignore those loudmouthed individuals.
On the other hand, publishing an API/protocol/whatever is applicable to your gizmo so that a 3rd party can actually add that Linux/Mac/Android/whatever support for the device is a good thing (e.g. if you are making a piece of test gear, having it working with Sigrok is always a good bonus).
Nobody can force you to support something you don't use and don't get paid for - just don't make it hard for others who would often do it even for free for you if the documentation was available. That's often the problem - most vendors make only a Windows driver for something and say they don't support anything else. Fair enough, but then they don't publish the interface/protocol of the device neither so that someone else can actually do the work without resorting to complex reverse engineering. And I am not talking about something super secret there where it could be kinda understandable but things like a wireless receiver for a gamepad (yay, Microsoft!).