Open source is not enough, since you have to trust the compiler and hardware the system is running on as well [1,2]. But from the page you linked, the key benefit of that system isn't that it's open source, but that the results are verifiable. I.e. you can check that the results are trustworthy, even if you don't trust the code or machines it's running on?

[1] https://www.win.tue.nl/~aeb/linux/hh/thompson/trust.html

[2] https://en.wikipedia.org/wiki/Intel_Management_Engine

I’m definitely more likely to trust open source. Generally the more eyes on it the better, I still think even one vulnerability could potentially be catastrophic though.

That microsoft link looks really interesting, I don’t understand automated theorem proving and crypto enough to see how it would work even with untrusted hardware (although it does say it can be run on any third party system)

What about voters' devices? Jessies WinXP, never updated, full of malware?