Revealing any private key for any type of Bitcoin wallet software kind of lets the person you revealed it to steal all your money.
I don't understand. I don't think you meant that non-deterministic wallet's keys have some kind of relationship that would allow revealing all the other private keys if any one of them is revealed, did you? And I wouldn't use only one private key in my wallet. More like one per transaction. Did you assume a single-key wallet when saying this?
I think that possibly the Bitcoin Magazine article has a mistake in it, or is maybe written in a way that gave you the wrong impression.
Did you have time to check out the article? If not, could you check it out, for example starting by searching for "crack_electrum_wallet(mpubkey,priv0,0)", a utility function Mr. Buterin mentions he has coded to demonstrate the issue?
Perhaps the subject of a hierarchical wallet is more interesting in this context, but if it applies to Armory's wallet, too, I still believe an arbitrary wallet user could easily think sharing a single private key with someone wouldn't add any risk, which actually would.
If you want to keep a backup for a gift to your friend, just keeping the private key won't be enough. You need the Chain Code too, as that determines all the addresses that the private key unlocks. The private key on it's own is like having a real-life key, but you don't know which lock it opens. When you know the house (think: address), and you have the key, then you can get through the door.
You are talking about deterministic wallets specifically, right? In case of Armory, I thought only the root key is something that fits all the wallet addresses. Do you mean that the rest of the private key chain can be figured out from any private address of the chain, even without the chain code? Or do you mean that Armory's wallet is hierarchical in such a way that not just the root key is something that works on multiple addresses, that any other private key had a similar property?