10 ms·

> "Users generally consider themselves better served if most unsolicited email they receive with no strong trust priors drops into a black hole." What is the e

> "Users generally consider themselves better served if most unsolicited email they receive with no strong trust priors drops into a black hole." What is the evidence for this claim? Not Microsoft or Google acknowledge they drop email.

shadowgovt 4y ago
Extrapolated from "Nearly 85% of all email is spam." At that volume, if your spam filter started with a random coin-toss you're more likely going to serve the user's interest than not. https://dataprot.net/statistics/spam-statistics/ https://dataprot.net/statistics/spam-statistics/
- pseudalopex 4y ago
  Conservative filters identify most spam. Rejecting email serves users' interests better than dropping it into a black hole.
  - shadowgovt 4y ago
    Rejecting email takes more than zero bytes back over the wire. When the other party in the transaction is assumed hostile, the principle is to provide them as little information as possible.
    - pseudalopex 4y ago
      The number of bytes transmitted back is not material to users' interests. Hostile parties can observe non delivery with accounts of their own.
- account42 4y ago
  That only follows if the cost of false positives and false negatives are the same, which they are not.
nottorp 4y ago
The fact that they don't acknowledge it doesn't mean they don't drop it. I've had to set up someone's mail server last year. All mails sent to gmail were silently dropped until we set up all the current buzzwords for the domain/email server. Then they magically started to show up. Possibly we were lucky that "just" setting up SPF DKIM etc fixed it.
- simplotek 4y ago
  > I've had to set up someone's mail server last year. All mails sent to gmail were silently dropped until we set up all the current buzzwords for the domain/email server. (...) Possibly we were lucky that "just" setting up SPF DKIM etc fixed it. SPF and DKIM are nowadays the very basic methods used to verify if your emails are spoofed or not. Not having them in place is as good as setting up a spam farm.
  - nottorp 4y ago
    Very nice, but we were talking about whether google silently drops emails or not. They do: the test emails were never rejected, but never arrived in the test gmail account either. Not in spam, not in all mail, not in the inbox.
    - simplotek 4y ago
      > They do: the test emails were never rejected, but never arrived in the test gmail account either. Not in spam, not in all mail, not in the inbox. And that's fine because that's exactly what SPF/DKIM were designed to do. The spam folder is not a dump of true positives. It's the bucket where you train the filter by evaluating somewhat likely false positives.
      - nottorp 4y ago
        What happened to rejection codes? I believe they're still defined somewhere in the SMTP standard? Or Google should be exempt from returning errors?

shadowgovt 4y ago

Extrapolated from "Nearly 85% of all email is spam." At that volume, if your spam filter started with a random coin-toss you're more likely going to serve the user's interest than not. https://dataprot.net/statistics/spam-statistics/ https://dataprot.net/statistics/spam-statistics/

pseudalopex 4y ago
Conservative filters identify most spam. Rejecting email serves users' interests better than dropping it into a black hole.
- shadowgovt 4y ago
  Rejecting email takes more than zero bytes back over the wire. When the other party in the transaction is assumed hostile, the principle is to provide them as little information as possible.
  - pseudalopex 4y ago
    The number of bytes transmitted back is not material to users' interests. Hostile parties can observe non delivery with accounts of their own.
account42 4y ago
That only follows if the cost of false positives and false negatives are the same, which they are not.

nottorp 4y ago

The fact that they don't acknowledge it doesn't mean they don't drop it. I've had to set up someone's mail server last year. All mails sent to gmail were silently dropped until we set up all the current buzzwords for the domain/email server. Then they magically started to show up. Possibly we were lucky that "just" setting up SPF DKIM etc fixed it.

simplotek 4y ago
> I've had to set up someone's mail server last year. All mails sent to gmail were silently dropped until we set up all the current buzzwords for the domain/email server. (...) Possibly we were lucky that "just" setting up SPF DKIM etc fixed it. SPF and DKIM are nowadays the very basic methods used to verify if your emails are spoofed or not. Not having them in place is as good as setting up a spam farm.
- nottorp 4y ago
  Very nice, but we were talking about whether google silently drops emails or not. They do: the test emails were never rejected, but never arrived in the test gmail account either. Not in spam, not in all mail, not in the inbox.
  - simplotek 4y ago
    > They do: the test emails were never rejected, but never arrived in the test gmail account either. Not in spam, not in all mail, not in the inbox. And that's fine because that's exactly what SPF/DKIM were designed to do. The spam folder is not a dump of true positives. It's the bucket where you train the filter by evaluating somewhat likely false positives.
    - nottorp 4y ago
      What happened to rejection codes? I believe they're still defined somewhere in the SMTP standard? Or Google should be exempt from returning errors?