3

Google has data available here on the proportion of emails outbound and inbound from Gmail using TLS encryption. Apart from an overall increasing trend, a striking feature of the inbound data in particular is a regular oscillation. Here is the past year's data for the proportion of encyrpted email arriving in Gmail inboxes:

enter image description here

Downloading the data and splitting it up by day, it's clear that the oscillation is a weekly variation coming from increased usage at the weekend:

Proportion of inbound TLS encryption by day

The vertical axis here is the proportion of emails using TLS, so 1 would mean all emails used it. The units on the horizontal axis are `week number', with week 1 beginning Wednesday 11th December 2013, when the data set begins (there's some days with missing data, which appear as the gaps in the graph).

There is some weekly variation in the outbound data too, but it's far less obvious, and doesn't seem to follow the same clear weekday/weekend pattern.

There is bound to be some degree of speculation here, but can anyone offer a plausible explanation? There are clearly factors affecting the volume and type of email sent like this (more work-related emailing during the week, or marketing emails targeted at weekends), but it isn't obvious to me why this particular pattern should result.

4
  • 2
    I don't quite get the dataset. Isn't it as simple as more emails being sent at weekends anyways? Jan 30 '17 at 17:40
  • 3
    I'm not sure how this is a security question?
    – schroeder
    Jan 30 '17 at 17:43
  • 4
    Could you please explain the axis on your visualization? Jan 30 '17 at 17:59
  • 2
    @RápliAndrás: No because the dataset contains the ratio TLS-mails/all-mails Jan 30 '17 at 18:50
2

I would need more elements to be sure, like the protocol used. What is clear when looking at the raw data is that the variation is mainly visible on inbound mail. That alone let me suspect that mails send through the webmail use TLS, while mails sent through SMTP relays do not. Which is somewhat coherent with SMTP protocol: it may go through a number of uncontrolled relays, so encrypting the transmission looks useless and many SMTP relay do not implement it. Instead they relay on the sender to encrypt the message content if privacy is needed.

That could explain why on work days, more mails are sent from work through corporate SMTP relays (no TLS), while less are on week end.


Below is my opinion:

Anyway, pretending that using TLS to and from GMail increases mail privacy is more marketing than information. What is the threat here? It is not far from securely closing a sheephold after letting the wolf in... The only way to increase privacy is to exchange encrypted content. But Google does not want to advise a method that would prevent itself to use the mail contents

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.