A System for Detecting Software Similarity
- Feb 1, 2018 And even more community contributions have been added!
- Nov 9, 2017 More community contributions have been added below ...
- Aug 31, 2017 Thanks to Chrstophe Troestler for an OCaml client for Moss.
- May 18, 2014 Community contributions (incuding a Windows submission GUI from Shane May, thanks!) are now in their own section on this page.
- May 14, 2014 And here is a Java version of the submission script. Thanks to Bjoern Zielke!
- May 2, 2014 Here is a PHP version of the submission script. Many thanks to Phillip Rehs!
- June 9, 2011 There were two outages over the last couple of days that lasted no more than a hour each (I think). I've made some changes to the disk management software that should prevent these problems from recurring.
- April 29, 2011 There was an outage lasting a few hours today, the first since last summer, but everything is back up.
- August 1, 2010 Everything is back to normal.
- July 27, 2010 The Moss server is back on line. There may be some more tuning and possibly downtime in the coming weeks, but any outages should be brief. New registrations are not yet working, but people with existing accounts can submit jobs.
- July 25, 2010 As many (many!) people have noticed, the Moss server has been down for all of July. Unfortunately the hardware failed while I was away on a trip. I am hopeful it will be back up within a few days.
What is Moss?
Moss (for a Measure Of Software Similarity) is an automatic system for determining
the similarity of programs. To date, the main
application of Moss has been in detecting plagiarism in programming
classes. Since its development in 1994, Moss has been very
effective in this role. The algorithm behind moss is a significant improvement over other
cheating detection algorithms (at least, over those known to us).
What is Moss Not? Moss is not a system for
completely automatically detecting plagiarism. Plagiarism is a satement that someone copied code
deliberately without attribution, and while Moss automatically detects program similarity, it
has no way of
knowing why codes are similar.
It is still up to a human to
go and look at the parts of the code that Moss highlights and make a
decision about whether there is plagiarism or not. One way of
thinking about what Moss provides is that it saves teachers and teaching staff a lot of time by
pointing out the parts of programs that are worth a more detailed
examination. But once someone has looked at those portions of the
programs, it shouldn't matter whether the suspect code was first discovered by Moss
or by a human; the case that there was plagiarism should stand on its own.
In particular, it is a misuse of Moss to rely solely on the similarity scores.
These scores are useful for judging the relative amount of matching between different
pairs of programs and for more easily seeing which pairs of programs stick out with
unusual amounts of matching. But the scores are certainly not a proof of plagiarism.
Someone must still look at the code.
Moss can currently analyze code written in the following languages:
Matlab, VHDL, Verilog, Spice, MIPS assembly, a8086 assembly,
a8086 assembly, HCL2.
An Internet Service
Moss is being provided as an Internet service.
The service has been designed to be very easy to use--you
supply a list of files to compare and Moss does the rest.
The current Moss submission script is for Linux.
In response to a query the Moss server produces HTML pages listing
pairs of programs with similar code. Moss also highlights
individual passages in programs that appear the same, making it easy
to quickly compare the files. Finally, Moss can automatically
eliminate matches to code that one expects to be shared (e.g.,
libraries or instructor-supplied code), thereby eliminating false
positives that arise from legitimate sharing of code.
Registering for Moss
Moss is being provided in the hope that it will benefit the
educational community. Moss is fast, easy to use, and free. In the
past, access has been restricted to instructors and staff of programming courses. This is no longer the case, and anyone may obtain a Moss account.
However, Moss is for non-commercial use. If you are interested in commercial uses
of Moss, contact Similix Corporation.
To obtain a Moss account, send a mail message to
The body of the message should appear exactly as follows:
where the last bit in italics is your email address.
If you already have an account, the latest submission script
can be downloaded here.
Security and Legal Stuff
If you use Moss, the results will contain copies of the code you
submitted and these will be accessible to anyone with the result URL.
Stanford can assume no liability whatsoever for your submissions.
Reasonable precautions have been taken to protect the confidentiality
of the code you submit. These measures include: only you receive the
result URL and it contains a random integer, so it is not easy to
guess a result URL. Furthermore, the directory with the results cannot
be broswed or indexed by robots. Finally, submissions are not
retained indefinitely on the server; typically results are deleted
after 14 days, though they may be deleted earlier to free up disk
space when the server is particularly busy. If some results you need have
been deleted, you can simply resubmit your job.
A number of Moss users have contributed versions of the submission script:
Andrew Cain has writtten a Ruby gem for Moss.
Hjalti Magnussion has written a summarization/visualization script.
How Does it Work?
A paper on the ideas behind Moss can be found here.
Back to Alex Aiken's Homepage