That's a bad argument. Firstly, it's my understanding that there have already been root-access 0 days discovered in the ME (and since patched since exposed). AND The USB jtag backdoor is the whole point of this post.
Secondly, a security hole and a backdoor are interchangeable these days. So we'll never be able to prove which new 0-days are deliberate, and as far as impact it kinda doesn't matter if they're deliberate.
You can't, but let me point out that DUAL_EC was a "nobody but us" backdoor that required their private key to use.
(and yes, it backfired)
If they're introducing regular vulnerabilities, they're also making themselves vulnerable, given that the US government is one of the biggest Intel customers.
I remember back in the good old days of cryptography export restrictions when the NSA had a much simpler "nobody but us" approach: you encrypted data with a xx-bit private key, half of which was shared with the NSA. Should they need to break content, the other half of the key could be brute-forced at costs that were economically feasible (for targeted use, not blanket suviellance) to the NSA but the full-length key would be unbreakable (in theory) by anyone without prior knowledge of that other half.
4chan has been popular for leaks/reverse engineering because of its anonymity and the fact that it's seen as (whether or not this is true, and I would wager that it isn't) as a "hacker haven". For example, a guy on 4chan reverse engineered Google's new captcha system almost as soon as it came out, leading Google to eventually hire him in exchange for his deleting the GitHub repo he was using.
While the ME is worrying for many reasons, there's absolutely zero evidence that the Intel ME contains a backdoor.
Backdoors don't stay hidden forever.