Simple Mail Email Service

[Tsunami]

Originally Posted by TenFour

^^This^^ Never a major problem in almost 20 years of using the same address. Best spam control in the business. My main complaint is their absolutely Byzantine security options and actions in practice. For example, yes they offer the ability to use a security key like the Yubikey, but they require you to also have a phone number and recovery email, as well as backup codes. And, in practice, I have found they often resort to sending an SMS to your account phone, regardless of what settings you have. I have also encountered times when they only offer one possible 2FA option, and if you click "try another way" they don't let you try another way. I almost lost access once when they would only send codes to an old phone number I no longer had access to and had previously been removed from my account.

On the Yubico website, Gmail is listed explicitly as a service supporting YubiKey.


Other than that, Gmail indeed is rock solid. Big tech or not, it's just a very good service.
And the idea that your emails are being read ... Maybe bots scan messages looking for keywords in order to send targeted ads. But that's different than an actual human being reading your private emails. So many sites use cookies, almost all social networks (BlueSky being an exception) send targeted ads... While that's not actually nice, it's bots doing all that.

Also, some services only collect metadata (where do you log in from, from which type of computer or smartphone, whom do you send messages to, ...) while they also cannot read any messages.

By this I, by no means, defend data mining as a way of income. Just saying that it's hard to totally escape from it, and that there are way more "evil" players around than Gmail.

[xylon]

Originally Posted by JeremyNicoll

Not in my view.

How do you unambiguously define precisely how a program is meant to work, to an AI? You could argue that the only way to do that is by giving the AI a program to read that already does it ...

Most people if writing instructions to the AI will not be precise / analytical enough in their natural language. Suppose the AI was trained on US English; will it also understand the exact same thing in other forms of idiomatic English?

How do you check what it produced actually does what you want?


Some of the problems I've seen include: code that would not work, won't compile, doesn't solve the stated problem but instead what the AI hoped you meant, mixes dialects of languages up.

Perhaps you are right.

[jeffpan]

Sending mail via Simple Mail is just simple.
https://ypeng.info/notes/send-simple...a-command-line