Analysing three months of academic spam

Can't get your paper in Nature? Publish it in the Journal of Ubiquitous Computing and Applications, just pay the $1000+ article processing fee, they accept everything. Your paper was rejected from the leading conference of your field? It had an acceptance rate of 0.5 % for the past twenty years, do not be surprised. Why not try the International Conference on Advanced Education Technology and Management Science? It is not your field, but they accept anything if you pay the $300 registration fee, and the proceedings will be indexed by the Whatever Junk Compendex.

Academic buccaneers try to cash in on high rejection rates, frustration, slow processing, and the high cost of open access publishing in renowned journals. They spam my inbox indiscriminately, ignoring my subject area. The calls for papers slip through the spam filters, no matter how much effort I put into training those filters. I collected the academic spam that arrived in my inbox between 2013-05-15 and 2013-08-30 to see what kind of observations I can make.

Academic spam in numbers

I received a total of 220 spam messages over the period, totalling 27 MBytes of wasted bandwidth. Out of the 27 MBytes, nearly 20 MBytes were from the same sender, a repeated email by the considerate organizers of the 3rd International Conference on Education and Education Management. The next worst offender sent an email only tenth of the size. Attachments otherwise were common, the most hilarious was when the zealous organizers enclosed the author instructions in a doc file. About 71 % of the spam messages had the mail in HTML format.

Total spam Unique addresses Unique domains
220 159 77

I processed the "From" field to extract the email address of the senders. Looking at the domains where the mails were sent from, there were only 77 of them (see the table above). The distribution was not uniform, the majority of the mails came from, more precisely, (see the figure below). The third worst offender was I am uncertain what grants a VIP status at these Chinese providers, but if it is for money, then it apparently enables the VIP address owners to spam indiscriminately. was also there, so not surprisingly the majority of the spam included passages in Chinese.

Pie chart of academic spammers

Proportion of the domains of origin of academic spam.

Google Mail came second, but this is misleading. Most of the Gmail messages arrived from the obnoxious organizers of the 1st International Conference on Advances in Computing, Communications and Informatics, in Mysore, India. This conference was borderline legitimate, with Vinton Cerf giving you the evil eye from the main website. Not all the blame goes to the spamming-prone organizers: EDAS provided the platform to send unsolicited calls in bulk. EDAS is a conference organizing tool. I registered two years ago when IPDPS was using it. Unfortunately EDAS has no respect for registered users, and allows seeking reviewers and sending calls to any user who ever signed up, even if you explicitly request not to. Moreover, it is not possible to delete an account. After a few email exchanges, this is the best that I could extort from EDAS tech support:

If you prefer to receive no email, including no invitations to serve on TPCs of any EDAS-managed conference, we can mark your email address as invalid and you will not receive additional email.

Since then, EDAS-related spam stopped, so I expect that spam from a Gmail address will stop.

About 68 mails originated from senders who built a crawler good enough to extract my name correctly. In four cases, they addressed my frequent co-author instead, but with the accents in his name displayed incorrectly. The others just started of with "Dear researchers", "Dear colleagues" (oh my, no!), or just "Dear", followed by a comma. Wolter Kluwer Health ambushed me with "Dear Medknow Member". I never wanted to be a Medknow Member at the first place. Eighteen papers included an apology for sending junk or potential cross-posting. A tiny fraction included a link to unsubscribe, but I never dared to click on it. That would be equivalent to writing a vexed reply to the usual "You inherited a billion dollars" scam.

A better filter

Filtering out any email that contains Chinese characters would solve most of my problems, but I maintain good relations with the Tsinghua University, and our correspondence frequently includes leftover Mandarin from other emails. Other sophisticated, content-based filters failed, so I need a simple, domain-based filter.

Since all my emails are in plain text format, they are easy to parse. An approved domain list is equally easy to create once I see the unique domain names. This script runs through my academic spam folder and generates a new regexp that includes all spammy domains (the sed in the third line should include a backslash before the parentheses, but it would be parsed as an equation by WordPress, hence the backslashes were stripped):

cat academic_spam/*|grep "^[Ff]rom: " | sed -e 's/[Ff]rom: //'|\
    sed -e 's/.*<(.*)>/\1/'|sed -e 's/ *//g'|sed -e 's/.*@//'|\
    sort|uniq > unique_domains.txt
echo -n ".*@.*(" > spammers.txt
grep -v -i -f approved.txt unique_domains.txt | tr '\n' '|' >> spammers.txt
echo "dummyemailaddress)" >> spammers.txt
rm unique_domains.txt

The current output is the following:


My domain uses CPanel, where user-level filtering allows using regexps. I set that emails matching this pattern should make their way directly to /dev/null. Waste of bandwidth is stopped at the server level.

Closing remarks

Are all open access journals scams? No, but the ones you receive a call from are. They are predatory publishers with fishy business models. Pieta Eklund wrote a guide how to identify the better ones, and Jeffrey Beall maintains a list of predatory publishers. The situation is far bleaker with conferences, and their calls make up the bulk of the spam. It is hardly enough that the organizers discredit themselves by being associated with such conferences. Most of these organizers are based in China, so it is also hard to take legal action. Academia strives on open communication, but your best line of defence is hiding your email address. I recently removed my email from all freely available PDFs of my papers, except the ones on arXiv, where no such changes are possible. This way at least an automated crawler will be less likely to pick up my address. For the spamming lists that already include me, the regexp generating script should reduce their annoyance.


  1. Brian Tiplady says:

    Thanks for a couple of really useful articles (this and the ACM one). I have the same issues with conferences and journals, and my only defence is specific message filters.

    I wonder what search engine you use? I've been trying DuckDuckGo and ixQuick, as much because of the declining usefulness of Google searches as the privacy issues. Both of these seem to replicate all the faults of Google (much as OpenOffice replicates the faults of Mocrosoft Office...)

    All the best -- B

    1. Peter Wittek says:

      As much as I hate to admit, Postini is a thousand times better at filtering spam than my manual tinkering.

      As for the search engine, I use Startpage, which is the same as ixQuick, as far as I understand. I find DuckDuckGo's interface without Javascript inferior -- and we do not want to enable Javascript, do we?

Leave a Reply

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

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>