Google Has Most of My Email Because It Has All of Yours

Republished by Slate. Translations available in French (Français), Spanish (Español), Chinese (中文)

For almost 15 years, I have run my own email server which I use for all of my non-work correspondence. I do so to keep autonomy, control, and privacy over my email and so that no big company has copies of all of my personal email.

A few years ago, I was surprised to find out that my friend Peter Eckersley — a very privacy conscious person who is Technology Projects Director at the EFF — used Gmail. I asked him why he would willingly give Google copies of all his email. Peter pointed out that if all of your friends use Gmail, Google has your email anyway. Any time I email somebody who uses Gmail — and anytime they email me — Google has that email.

Since our conversation, I have often wondered just how much of my email Google really has. This weekend, I wrote a small program to go through all the email I have kept in my personal inbox since April 2004 (when Gmail was started) to find out.

One challenge with answering the question is that many people, like Peter, use Gmail to read, compose, and send email but they configure Gmail to send email from a “From” address. To catch these, my program looks through each message’s headers that record which computers handled the message on its way to my server and to pick out messages that have traveled through,, or Although I usually filter them, my personal mailbox contains emails sent through a number of mailing lists. Since these mailing lists often “hide” the true provenance of a message, I exclude all messages that are marked as coming from lists using the (usually invisible) “Precedence” header.

The following graph shows the numbers of emails in my personal inbox each week in red and the subset from Google in blue. Because the number of emails I receive week-to-week tends to vary quite a bit, I’ve included a LOESS “smoother” which shows a moving average over several weeks.

Emails, total and from GMail, over timeFrom eyeballing the graph, the answer to seems to be that, although it varies, about a third of the email in my inbox comes from Google!

Keep in mind that this is all of my personal email and includes automatic and computer generated mail from banks and retailers, etc. Although it is true that Google doesn’t have these messages, it suggests that the proportion of my truly “personal” email that comes via Google is probably much higher.

I would also like to know how much of the email I send goes to Google. I can do this by looking at emails in my inbox that I have replied to. This works if I am willing to assume that if I reply to an email sent from Google, it ends up back at Google. In some ways, doing this addresses the problem with the emails from retailers and banks since I am very unlikely to reply to those emails. In this sense, it also reflects a measure of more truly personal email.

I’ve broken down the proportions of emails I received that come from Google in the graph below for all email (top) and for emails I have replied to (bottom). In the graphs, the size of the dots represents the total number of emails counted to make that proportion. Once again, I’ve included the LOESS moving average.

Proportion of emails from GMail over timeThe answer is surprisingly large. Despite the fact that I spend hundreds of dollars a year and hours of work to host my own email server, Google has about half of my personal email! Last year, Google delivered 57% of the emails in my inbox that I replied to. They have delivered more than a third of all the email I’ve replied to every year since 2006 and more than half since 2010. On the upside, there is some indication that the proportion is going down. So far this year, only 51% of the emails I’ve replied to arrived from Google.

The numbers are higher than I imagined and reflect somewhat depressing news. They show how it’s complicated to think about privacy and autonomy for communication between parties. I’m not sure what to do except encourage others to consider, in the wake of the Snowden revelations and everything else, whether you really want Google to have all your email. And half of mine.

If you want to run the analysis on your own, you’re welcome to the Python and R code I used to produce the numbers and graphs.

229 Replies to “Google Has Most of My Email Because It Has All of Yours”

  1. I am in a similiar situation. The thing that frustrates me in addition to this is the total lack of adoption of PGP/GPG. Better privacy is so close and hardly anyone will make the effort.

    1. Make it easier for the end “non-technical” users to adopt GPG/other technology in a seamless way, and things will flow automatically.

      1. Bullfeathers! Using GPG is no more difficult than learning how to play Angry Birds, and a lot simpler than configuring Facebook privacy settings.

        The people who complain about the difficulty of GPG seem to be people who have never tried to use it.

        Can you log onto a website? You can use gpg.

        1. Using encryption is not just about pushing the buttons but really requires understanding the concepts. I’ve seen people using OTR and GPG, and that don’t really understand them, do things like send me their private keys! There are attempts to make encryption easier to use. One I have been excited about recently in regards to email is the LEAP project.

        2. “Bullfeathers”?? Such a nice, arrogant attitude. And comparing it to Facebook only makes your case WORSE! Maybe YOU have time to struggle thru confusing software, but *I* do not.

          I use PGP but it is clumsy, requiring a significant amount of extra time and work for every email that I read or write that is encrypted.

          So I tried adding GPG to my Thunderbird–and it FAILED. I was NOT impressed–and I damn well don’t want to spend MORE time debugging bad code written by someone else.

          I am impressed with Bitmessage, but it is also lacking simple, rational installation procedures. However, it is currently Beta, so there is hope that the developers will have a better attitude about “User Friendly” installation.

        3. Well, generating the keys and keeping them synced between your devices is difficult.

          If you loose the keys or access your email from a device where you don’t have the keys (your new phone, laptop, a friend’s computer) you are not going to find GPG user-friendly. People have more than one computer and expect to be able to access their emails everywhere.

          1. People who are true geeks find some things easy that may as well be written in Chinese for the rest of us. Because it is easy FOR THEM they refuse to listen to honest feedback from people like me.

            It is the same reason so few use Linux. All they need is for commonly used programs to have autoinstallers – but geeks don’t need them so they refuse to be bothered. They don’t understand that without them only a very small percentage can use Linux.

            And that is the same reason most people don’t encrypt anything or us proxies.

      2. I’m a “technical” user… and I’ve tried several time to use GPG and S/MIME… it’s horrible to use.
        No nice and easy to use plugins available for most mail client. Bad gpg tools which prevent to use a long and strong password in softwares like keepass.
        No real standard on how to send crypted attachments with emails… so no usable with differents people using different mail client.

        1. just use OTR. pidgin, jitsi and plenty of other clients support it and it requires zero technical knowledge.
          wtf man.

    2. But gmail can also hadle PGP/GPG through extensions if I’m not mistaken. If so, maybe encryption doesn’t help for hiding emails from google?

      1. Doesn’t matter if gmail has PGP extensions. By the way, I think it was possible but was ditched recently (I can guess why:)). Anyhow, whether they can use PGP or not is totally irrelevant as long as they don’t compromise your private key. So one can easily and safely send email over google servers with PGP. Also, there is a solution which I find quite neat – if you write to non-pgp-concious people, you can divide your message to parts which you don’t give much about (non-private) and that “private parts” (hehe), and encrypt just “private parts”. This way, if they want to decrypt, they can. If they have a key of course.

        One more thing: unencrypted email is like a postcard. So the question whether or not google has it is less pressing than whether or not ANYONE has it, because, if PRISM is connected to internet hubs… well, how to put it. Guess yourselves:)

        1. It might be worse than you think. I use my personal domain to send / receive messages, but I auto-forward everything to gmail as they offer large storage space.

      1. As it has been mentioned, your blog served fonts from Google.

        Not to point fingers. Just illustrating something which is about the friction and the efforts. In between two parties interacting, communicating, there is a way to understand the cost of participating for knowing where is the power. Right now, when we want to be privacy conscious, it takes each of us, individuals, a lot of efforts to do so. And that is the issue, the cost of non-interacting has become higher than the cost of interacting, and it’s why many people are engaged to do things even against their own beliefs.

        1. First, thanks for pointing out the font thing. It seems to have come in with a theme I installed recently and I did not realize it. Removing the font from the theme causes it render incorrectly so I’ll need to look into a better solution.

          That said, the situation is a little different. For one, I am under the impression that Google claims they do not store or monitor traffic to their API servers for more than a very short period of time and they do not correlate it with your Google account. It uses a different domain (not unassociated with your Google cookie .

          Second, you can view my blog without loading the font from GoogleAPIs. It will be a little uglier without the web fonts but it will work find otherwise. In this way, I unilaterally block many advertising and tracking networks (including several from Google) and I can visit websites just fine. On the other hand, I can’t send or receive email from a Gmail user without Google having a copy of that mail.

          Third, and most importantly, for me I don’t care nearly as much about privacy in my web viewing data as I do about my email. I do care some, but less.

          In at least one sense though, the situation is worse. There’s no reason for you all to know that you were visiting Google when you visited my site. Except, as others have pointed out, that most webpages send your data there, — usually in ways not nearly as benign as a web font.

          1. As a web developer, I have to agree that you should host your own fonts, especially if you have gone through the trouble of using piwik instead of google analytics for exactly the same privacy reasons.

            Here’s a little guide:

            You can get the Noto family of fonts to self host here:
            The Inconsolata console font you can grab from your local workstation.

            Font squirrel will help you make a webfont stack with them which you can drop back into your theme. You will need to wire up the css font family names, which font squirrel’s example code will help you with

            Additonally, you may need to tell your webserver to serve the woff mimetype.

            Good luck!

    3. The reason nobody bothers with increased security is because it has no impact on their lives, it might creep people out that Gmail uses a bot to read emails for keywords but again it has no impact on their lives or security.

      1. Agreed. Whether Google has or does not have some/all of my email is related to what they plan to do with it, and we already know it’s to serve ads. Which is relatively benign. And mostly only applies to email in and out of Google.

        Much more of a concern are the government-level packet captures running on ALL email traffic and ALL web traffic, and those agencies aren’t interested in serving up ads. Whatever goal that have in mind is likely to be far more at odds with your privacy and security than anything Google has in mind. Running your own mail server won’t help. Running Python scripts won’t matter. They have it all.

        So I don’t worry so much about Google. Privacy is an illusion.

    4. Actually one of the reasons to choose gmail instead of other ecternal providers was that it gives you some encription and cloaking agains minor players (private detectives, company bridges etc)

  2. “dig -t mx” will tell you whether a given domain gets its email handled by google. If you do that with the addresses you send messages to, the results are not nice. You might try running a script with this test over any outbound messages that you have archived.

    1. This will only tell you the mail server the domain currently uses, not the one it used at the point in time when the mail was handled.

      I like the people best that hand over their entire mail to some random webservice provider by giving them their login credentials just so that the webservice provider can connect one to/or invite other people one knows.

      1. Well, I looked at the “Received” headers so I will see any server that is touched it in a way that is recorded in those headers. I understand that my method will not catch everything but I think that makes it a conservative measure.

      2. Yeah, “I like the people best that hand over their entire mail to some random webservice provider”, those people really are the best.

        And that’s why I decided to set up my own server with all the necessary tools. I even thought about only allowing encrypted server-to-server-communication, but that would deny about 80% of all messages of being sent/received.

  3. Verifying a suspicion with actual data, kudos!

    I agree with John, sending emails to/from Gmail with PGP/GPG would be the best outcome, even if using Gmail.

    I propose we encourage widespread PGP/GPG adoption by competitions to win prizes, simply email in asking to win with PGP/GPG and you’re in the chance to win.

    Or perhaps make mobile phone apps that allows you to exchange naked photos safely using GPG/PGP.

    1. I use GPG often. That said, the number of emails I send to GPG is a tiny fraction of the total because most of the people I correspond with don’t use it. The number of GMail users I correspond to that use GPG? Perhaps none at all.

      1. I’d like to use GPG more but privacy ultimately depends on the other person using free software. The tools exist in free software. Kmail, Evolution and other clients are relatively easy to use. I understand the difference between private and public keys but don’t quite understand the finer points of the web of trust. I have not figured out automatic key import and I’m not sure I can trust it because the NSA broke the cert system. There are two or three people who I’ve exchanged public keys with who are smart enough to not use them on a non free software compromised computer.

        It’s an uphill battle with many well fortified positions. Wintel is putting UEFI on people’s desktops, so even though people can buy a computer with nothing but free software, it’s got hardware backdoors. Cell phones are even nastier – useless for private communications and so invasive you should not even talk around them.

    2. This would do nothing about the metadata issue.

      Thank you for this analysis, Mako!

      P.S. I am in the same situation as your friend. Gmail because they have my email anyway…

  4. Out of curiosity, have you tried repeating your analysis on the other major webmail providers run by US companies? Hotmail, Yahoo, etc.

      1. So, you might consider changing the title of your article. This is not a Google issue but an email issue. Of course, Microsoft and Yahoo and every other such service has your email, too.

        1. Well, having one company with most of my email worries me because of its particular potential for abuse — even if other companies have a lot of the rest. Since in my case, Google has the most of any single company, they are the biggest problem in this sense.

          1. Isn’t the key then to treat email like a public Facebook wall and watch what you say? No more sharing information via email you wouldn’t otherwise shout out from the public square.

  5. This is why I plan on configuring my mail server to bounce all email coming from corporate surveillance whore sources.

    If my friends want to use email to communicate with me, they’ll have to get with the program and ditch the PRISM providers.

    1. Radical solution. Do you still get emails or did your friends decide it wasn’t worth the effort to mail you any longer?

      1. I haven’t implemented it just yet. I’m going to custom tailer the bounce message to communicate my reasons as well as include links to more dignified providers and solutions.

        But if it’s 2014 and they still don’t want to give up these services, which I believe are harmful, then too bad for everyone I guess.

    2. What difference does it make when the emails are sucked into NSA central? Unless you PGP/GPG the lot. And even then, email leaks meta-data.

      Email is unprivacy. Stop using it.

  6. I wish some large email provider, such as GMail or Yahoo Mail, would start using end-to-end encryption routinely, and transparently. When you click the Send button, software (maybe an open-source browser plug-in) looks to see if your recipient has a preferred encryption method and public key registered anywhere (or if one is cached locally, via prior key-exchange). If recipient does, the message gets encrypted (by open-source browser plug-in) via that method before sending. If recipient is not registered anywhere, message goes unencrypted, as usual. Simple ! And now the email provider itself can’t read or decrypt your messages.

    This would be transparent to the sender, and could be transparent to the recipient. Recipient just registers a public key and method in a public registry somewhere. Private key is on their machine.

    It is perfectly feasible, technically, for Google, Yahoo etc to do end-to-end encrypted email AND still do targeted advertising. Do encryption via browser plug-in (maybe open-source), and have the plug-in extract a few keywords before encrypting. Send the keywords along to server with the encrypted message, so server can do advertising.

    1. Doing that would mostly kill webmail. (Even if the key is stored locally, it means you can no longer access your email from another device.) Unless the email provider has the secret key, in which case there’s very little point (since the providers at both ends can read the email contents, and assuming the point-to-point encryption works correctly no one else can anyway).

  7. Strange reasoning: “Google has our email anyway, so I use GMail as well.” It’s like suicide for fear of death. Please, everybody, use decentral, local email providers! I do :~)

    1. Exactly me first thinking. I mean, that’s really a ridiculous argument:

      “Peter pointed out that if all of your friends use Gmail, Google has your email anyway. Any time I email somebody who uses Gmail — and anytime they email me — Google has that email.”

      Eh, and for that reason you add “+1” to the number and make sure everybody who emails you also emails Google? Effectively, you are agreeing that it’s good that everybody uses Google. The state of EFF is poorer than I expected…

      1. I don’t want to put words into Peters mouth but it’s easy for me to see his reasoning and I don’t think it is ridiculous. There are lots of positives to using Gmail. It is also cheaper and easier than running your own server and perhaps it has features or an interface he likes. The fact that Google has his email anyway simply weakens the privacy objection that is weighing on the other side of the scale.

  8. Btw. GnuPG is not a sufficient answer to the privacy problem: It does not hide the meta data (who is writing to whom, when, how often, how much), which is in most cases more important then the actual content.

    1. @Martin
      “It does not hide the meta data (who is writing to whom, when, how often, how much), which is in most cases more important then the actual content.”

      Keep watching BitMessage. They are working on just those issues.

      I have collected some info at “BITMAIL anyone?”

  9. Your isp has all of your unencrypted (they are such if either you or the recipient do not support encryption, so pretty much most of them, regardless if google has ever touched them) emails.

    1. Not true if you communicate with your email provider with SSL/STARTLS/some other encrypted means. Your emails, if unencrypted with pgp/gpg, will be accessible by your provider, but NOT your ISP (or any other man in the middle for that matter, unless they’re using SSLSNiff, Dsniff, or some other tool to intercept the key exchange).

  10. Great data analysis, thanks!

    It would also be great if others with such a long data series would like to check and publish their findings if they have the same trends.

    Would also be very interesting and educational to check Hotmail, Yahoo etc. It will probalbly show the propotions between the major mail service providers, esp. since you have such a nice long data serie to work with.

    1. If you do run the scripts, post a comment here with a link to your analysis. I’m interested in seeing how my numbers stack up to other peoples. I am helping somebody fix some issues as they try to run the scripts right now and improving the documentation a little bit for people who don’t know R.

  11. Does this include people that use Google Apps? It might be even higher.

    If your email correspondence is sensitive encrypt it. If you need to protect metadata then maybe email isn’t the best communication method.

    1. I do encrypt my email but only to other people that use PGP/GPG. That’s a small proportion of the people I communicate with. And because I know lots of people use that crypto, I am probably in a better situation in this regards than most others!

  12. What if we send our messages in a harder to read format, like in images? An email could contain one large image or an image for about each word. If done correctly -no white on yellow for example – almost anyone will be able to read the message without special software to decode it. You can make the message more visually appealing, or at least interesting, than text or a web page and bypass any email text processing programs. I am assuming that all images aren’t OCRed by default already.

  13. Most mailinglists shall have one or more subscribers with a Gmail-address. So Google has most of the malinglists-posts.

  14. “Peter pointed out that if all of your friends use GMail, Google has your email anyway”

    ARRRRRRGH! This is absolutely *no* reason to do it yourself too!

    I’m hosting my eMail within my own appartement, TYVM!

    Thanks mako for your insightful writings.

  15. The worst part of it is that your e-mails to everyone (non-gmail) else use servers, and are on the internet.
    Your ISP has access to everything else, since you are connecting to the internet through them.

    And now with facial recognition being so advanced when you go to London, or Chicago, they can follow your face in a crowd and say “Hey, that is the guy that was e-mailing Youtube videos of Nyan Cat and Obiwan Kenobi!”

  16. Also, a high proportion of email consists of contributions to long conversation threads, with most of the thread preserved but hidden in clients by default, and that gets smashed all over the Internet because every participant has a copy.

    Which makes the proportion that’s out of your control even higher. Sigh.

    BTW, people who are interested in making PGP/GPG more usable by ordinary people might want to check out

  17. Mako, being the extremely smart social scientist that you are, you should point out the selection issue — your friends are likely to be at least half as anal as you are (well, the ones not in business school, anyway) and are therefore *less* likely to be using Gmail. I should send you metadata from *my* email (or run your scripts myself) — I’m quite sure for the “average” person, the numbers would much higher — closer to 90%. I think the one thing that saves people is “work” email, when you communicate with official email addresses, but there the challenge is knowing if the email is hosted on Google Apps.

    1. This a very good point, Abhishek. Thanks for making it here! This analysis is about my email and there are many, many, reasons to believe it’s not other peoples email. I would expect, among my group of friends, that you’re right and that my numbers are actually quite low relative to most people I know. I suspect a large majority of my friends email never leaves the Google network.

      If you are comfortable with python and R and can get your data into a Maildir or modify a Mython program, go ahead and run the analysis! I would love to see the results to see how your experience stacks up.

      I did not include my “work” email in this (i.e., my email from MIT or UW or Canonical which barely overlapped with the beginning of this period). I am quite sure that the proportion of Google messages in my work email is much lower. That said, I don’t get choose my work email provider. This article is about my (somewhat futile) attempt to control my own email by running my own server. As a result, I only included the inbox that is impacted by that decision.

  18. I will continue to run my own mail server and I have been teaching my friends how to run one as well. Also, if you run a mail server on a rental server or vps, be sure to set up encrypted partitions for your mail, your certs and other config files.

    1. I will also continue to run my own server. Nobody should interpret my post as trying to talk people out of running their own server. It’s just that, today, I’m feeling a little futile about the whole thing.

      1. It’s not futile. Having a server is not only for you, but for the other people too:

        If some of you friends wants to improve his privacy, then he can rely on you if you’re using GPG, your own server, etc.

        With you script, I got 25% of all my email that goes to google, but 75% for the email with replies…

  19. Even though GMail may have a significant portion of my emails, they don’t have it in one place. If presented with a request for all of my emails, they couldn’t supply anyone with my inbox. It is doubtful, that they would be forced to scan everyone’s email box for messages that include me.

    Google may have more information about me than I like, but I can do my part to keep that to a minimum.

    Great article. I never thought about it before.

    1. You think so? I would guess that they do in fact store all the emails in on place and that separate inboxes are really just a fiction in the user-facing code.

      As in, do you really think that if I send an email to 1000 GMail users that Google creates 1000 copies of that email? I guess they just store one copy and 1000 references to it in their database. That should be much faster and more efficient and Google has lots of reasons to care about those things. Why wouldn’t they do it this way?

  20. Email that you send out to another person is no longer your private email. It’s more theirs than yours.

    1. That is exactly the same problem as having your email under GMail… it’s not under your control.

      I would say is worse since probably neither or hire first-class hackers to keep their sites safe.

  21. Email privacy is an oxymoron. I still laugh when I see fax era confidentiality notices appended to emails.

  22. I think that encryption is an important, but only partial solution, as others have noted.

    I think that making visible that this is the way email works (like this post does) is also important; that helps to build up norms (which already exist) and laws (which, as usual, lag behind) about what people and companies can/should do with our data.

  23. Pingback: Anonymous
  24. As increasing infringements on our Email and online privacy rises, we see great demand for a solution. The threats against your personal Internet privacy is increasing everyday as “free” Email providers, hackers, NSA’s PRISM program, and the amended US Patriot Act are just a few of a growing list that are compromising our freedoms. As we stand at a crossroads, it may appear hopeless to protect our God given rights to privacy but rest assured, there are real solutions to this serious problem! offers 100% guaranteed online privacy because our servers are located in Switzerland, a safe-haven for secure digital communications. As a law abiding citizen, you can be sure your digital data is safe from any agency, business, or anyone at all wanting to retrieve your information. Access to your online data communications by any authority requires an official warrant issued by a federal judge of Switzerland while most countries surrender your data without consent.
    To further protect your privacy, we delete the ‘Received’-Header that contains the customers IP & All incoming emails are scanned for viruses and drop all infected emails.
    Our VPN service changes your IP address every 10 minutes and our DigitalSafe is a “Swiss Bank” for your data! DigitalSafe has a feature whereas you may send a secure note that requires a PIN (password) in order to view it. It routes the unsecured email user to a secure site to read the encrypted message. They may not respond as they have read only rights. Visit my site and feel free to contact me with any and all questions you may have!
    There is secure email, and then there is Swiss Secure E-Mail…

  25. …. and there’s probably a bunch of “forwards” and “reply-to-all’s” of your original emails from within your friends’ gmail accounts

  26. You can frustrate google’s automated analysis of your email by using a custom address for each person you correspond with. For example, my mom thinks my email address is:

    It would be possible for google to cross-reference all addresses at a domain, but they are unlikely to because determining what is a single-user private domain and what is a multi-user domain takes a human to get right.

  27. Mako, don’t take this personal, but I blame you programmer types. “sheeple” out there use Gmail because you programmers haven’t come up with something better. It is possible to create mail, social networks, instant messaging, VOIP apps, that all communicate peer to peer without a central hub. The coming IPv6 switch-over will make this more feasible than ever.

    If you build it, WE, the Alpha nerds of our friends and communities will make sure our flocks will use it.

    So stop telling us what we already know, and DO something about it already! :)

    1. Well, my current project is not around building systems either. I want to understand and communicate about problems we’re facing and do everything I can to more people learn how to program and change their technology infrastructure.

      I think that that the problem is that there aren’t enough programming types out there out there who are interested in solving this problem! Perhaps you want to start learning to program and to join the effort?

      The burden falls on all of us that care to do whatever we can.

  28. A while ago I had made the suggestion that one solution would be for the sender to not send the mail but only send a link to the mail. The mail would actually be stored by the sender on a secure server. Most modern (web)mail readers would automatically open the link. Authentication for the server could be provided using a number of mechanisms including Google. This way mail is kept private.

  29. The first problem is not that people don’t use gpg or some kind of other encryption like otr when chatting or use their own mail server. The problem is in the education of privacy and the problems/implications of losing privacy or having no privacy.
    Most (non technical minded) people don’t see it as a problem when losing privacy. People want to share and communicate with other people and don’t see the danger when privacy is compromised by a government or some company. Otherwise facebook and google would be out of business.
    And I think this is where the discussion has to start. For example the whole Snowden revelations has sparked a discussion in certain circles, but at the end most people don’t see the danger or don’t care, because their lives revolve around other things then privacy and all it’s implications. This is where it has to start and it’s a really hard problem to tackle. And I don’t say that an easy to use implementation of encryption which is transparent to the end user won’t work, because it will start to make a difference, but it’s only one aspect of the whole picture. And of-course blog post like this also help. But there has to be a ‘easy’ way to start people educating about privacy and the danger of having none.

  30. I personally recommend openmailbox for regular mail, they run on free software exclusively. Hosting a personal mailserver is also an alternative, virtual servers are really cheap nowadays if your ISP doesn’t allow mail servers. However, we should really try to adopt Bitmessage or some other, more secure way of communication.

  31. Surely this is an underestimate. Many people with their own domain name redirect to/from gmail. I run my own domain and email server but I back everything up to gmail so when you email me my server transparently sends a copy to gmail so that I have an easily searchable back up. There’s no way you could know that or tell that from any headers you see.

    1. I am only looking at the headers of emails that I receive only. If you forward email to Gmail and respond using Gmail, I see that and will count you as using Gmail. If you use Gmail as a pure backup — i.e., sending and receiving mail from your own mail server but forwarding an additional copy to Google’s server — I will not be able to count that as Gmail. My sense is that this is not extremely common.

      Much more common I think, and also missed by my scripts, is CCing. If somebody sends an email from a non-Gmail address but CC or BCC’s a single person who uses Google as their mailserver, Google has a copy of the email and I have no way to tell.

      You are right. For these and other reasons, my estimates are a lower bound. The real answer must be at least a little higher.

  32. Obviously. May I suggest yet another obvious article about two plus two equals four? What a waste!

    1. It was obvious to you that I receive 57% of my email last year from Google and that after nearly a decade of steady increase it now seems to be going down? Was it also obvious to you that others who have run the scripts have values as low as 15% but (so far) usually higher than mine and as high as 80%?

  33. Pingback: Google has my mail
  34. Nice trivia, that i have to enter a valid E-Mail to be able to reply to this entry. Anyway:
    I have to accounts on a private server: I checked
    All Mails: 576.819 – with 72.698 Hits = 12,60%

    Private Mails: 419.172 – with 13.112 Hits = 3,13%
    Mails as a member of Pirateparty standing for privacy
    : 157.647 Mails – with 59.586 Hits = 37,80%
    Trash-Folder: 112.315 – with 18 Hits = 0,02%

    Well, i conclude my Trashmail is almost private. ;-)

    Thanx for the article.

  35. Hi,

    is there a way to make this statistic with a thunderbird-profile with its stored mails?

  36. Isn’t it still a difference whether you have allowed Google to use your data versus not having allowed that if only your correspondent has a Google account?

  37. This is the point: To have your own server, you spend +100bucks and a ton of work. Thats the whole thing why everyone uses gmail. I hope for a day when someone simplifies the process of obtaining a tiny dedicated server (like a raspberry) and setting up a stable mailserver on it. Simplyfies that much so the instructions are ready to casual computer users. That will be gmails dead end :)

  38. …the fact that I spend hundreds of dollars a year and hours of work to host my own email server, …
    What makes this so expensive? Doesn’t seem very compatible with ideas of easily/cheaply hosting for yourself, like Freedombox…

    1. If you want to host a server on a reliable Internet connection (i.e., at an ISP and not at your home), it will cost hundreds. Well, $200/year is under $17/month. That’s about the price of a reasonably slow virtual server. I had to upgrade to a more expensive virtual server in order to spam check all of the email I receive.

          1. Aaron Swartz on RimuHosting.
            It’s an article from 2009, but I thought you should read it.

          2. I read Aaron’s article when he wrote it. I was probably even one of the people who he described recommending Rimuhosting to him! I’ve been using Rimuhosting for decades and I have only had wonderful experiences. I’m not sure why Aaron seemed to have a such a rough but I can say that my partner Mika and I both use them and have only had great experiences. Despite Aaron’s experience, I would strongly recommend Rimuhosting to anybody.

  39. Thanks very much for this — this is a great tool. I finally got around to running this on a series of mailboxes. For those who don’t keep all their mail in a single mailbox, you can do something like this:
    $ for MAILBOX in ~/Maildir/.people* ; do python $MAILBOX >> mail_metadata.tsv ; done
    I’m seeing a continuous increase of email coming from Google, just over 50% overall, and just under 50% of email with replies.

  40. Pingback: Fefes Blog
  41. Pingback: Sofs
  42. I don’t understand the problem. Goggle has your emails, so what ? You’re afraid by what exactly ? Google reading your private conversation ? You (and all the people) speak about Google like it’s a person. No it’s not. Google is a company. Google has billions mails. Google doesn’t care that you sent “I love you” at your girlfriend the third November 2004.

    Google has the half of your personal emails. And the other half ? Microsoft or Yahoo have them I suppose. Maybe Google is meaner than the others… I don’t think so. Therefore if I follow your logic, everybody should have his proper mail server. And we will be sure that no big company have our emails… Even if, technically, nothing prevents the ISP or other carriers in the chain from copying and keeping your email.

    1. You can tell if sending to a given address goes to google, by looking at the MX record of the target domain. So for you’d say “dig -t mx” and look for google servers in the response.

      1. Lots of people forward their email. For example, many people at the University of Washington ( use GMail to store, read, and send email but the primary MX records in DNS point only to UW. For users that use GMail — either as an official interface offered by the university or simply by forwarding their mail — an MX wouldn’t show this because all the mail goes through UW on the way to Google. Looking at the messages they send in the way I did, however, would reveal this clearly. It’s not a perfect answer, by any means, but it’s best I could come up with.

  43. With respect to the self hosted email server, this is not a solution to that can offered to everyone, not because of price, but because of the hassle.

    There are actually too many services out there claiming end-to-end encryption like Protonmail and Tutanota. What do you think? what are your suggestions?

    1. I like the model of LEAP which is for many federated systems. The issue is not that everybody needs to run a mail server but because they can use infrastructure run by an organization or person they trust instead of one big giant profit-driven machine without our best interests in mind.

      1. While I’m using my own server for my emails, I can’t agree with this. For an ordinary person it’s much more creepy if your sysadmin friend will read your emails than if some unknown person or algorithm at Google reads it.

        Your friends or acquaintances are more likely to be interested in your email than a corporation. Not everyone knows a technical person they can trust not to read their stuff.

  44. Suppose it’s true. There is still the advantage of Google not having it in one place and needing to mine their data to find anything of use. I would imagine they’re not going to trying to data mine information about you from everyone else, because why bother?

  45. Holy shit. I can’t believe Peter Eckersley has such a whimsical attitude about using Gmail. His job apparently:

    > He leads a team of technologists who watch for
    > technologies that, by accident or design, pose a
    > risk to computer users’ freedoms—and then look
    > for ways to fix them.

    I guess being technically minded, Eckersley doesn’t recognize or can’t compute the risks to user’s freedom which information monopolies represent. Obviously he knows it, but apparently it means nothing to him.

    I don’t mean to bash him. But I would expect that if we had to choose between promoting insecurity through using email (and Gmail at that!) and using other, more secure communications technologies, that an EFF heavy would choose the latter, and refuse the former. Otherwise, what is privacy advocacy?

    1. I’m sure that Peter does use alternatives to email and he certainly uses GPG. Like most people, he also interacts with people that do not yet use those other “better” systems and his ability to do his job and run an effective organization would be severely hindered, to say the least, if they refused to use email. We make trade-offs. In this case, Peter’s reasoning seemed to be largely supported by the facts.

  46. I was emailing a Gmail-using friend today and found myself deleting about 1/2 of it before sending.

    This quote from the comments sums it up: “Eckersley doesn’t recognize or can’t compute the risks to user’s freedom which information monopolies represent.”

    I stop by this post now and then to see the comments and forward the link to a friend. I also check if the fonts are still coming from Google. They are. Good luck with that.

  47. Hello Mako.
    Can you please tell your email stack? I’ve heard so many opinions on the available options and would like to add yours ;)
    greetings, a person

  48. Several years later your original post do you have any update? I would love to discover what’s that percentage nowdays.


    1. Same here. I check MX records before emailing people and have a filter warning of inbound. I simply refust to email Gmail users.
      I am noticing that some are moving to FastMail, ProtonMail, etc.
      Nice to see.

  49. I am really not okay with Google anything. Sorry to see the Google fonts and thus tracking on this blog. Oh well. It boils down to a lack of respect for user’s privacy or freedoms.

  50. lolol – 2 wrongs dont (EVER) make a right… Google is evil period. My family uses Protonmail, so until Google buys them, too, they don’t have our communications. You are an idiot to just give up.

  51. The convenience of Google services often comes at the cost of our digital footprint. Your perspective sheds light on the reality that while we willingly engage with these tools for productivity, there’s a tradeoff in terms of privacy. It’s a crucial conversation we need to keep having

  52. This is such a thought-provoking perspective! It’s amazing how interconnected our online presence is. It’s like Google holds a piece of everyone’s digital life through email. Makes you stop and ponder about data security and privacy in this hyper-connected world.

Leave a Reply

Your email address will not be published. Required fields are marked *