Metadata and Academic Honesty
February 4, 2016 10:44 AM   Subscribe

I'm teaching an undergraduate photography class this semester. I require that each assignment be photographed within the assignment period. I have a (very talented) student who twice has submitted photos he had in his archive rather than taking new images for the assignment--confirmed by checking the metadata in his images. Is there any chance that I'm wrong about this?

He admitted to using old images for his first assignment and confirmed that the dates in the metadata were correct. I graded him harshly and let him know if it happened again he would be receiving no credit for the assignment. When I questioned him on the second assignment, he insisted the images were taken over the weekend, but the metadata suggests they were taken in 2013 (with the same camera/serial as the other images where he confirmed the metadata was correct).

Is there any way I'm wrong about metadata being (mostly) infallible? Is there a scenario in which I could believe that the metadata was mis-read or mis-written in to the file? I'm almost 100% sure the images were not shot during the assignment period, but I'm frustrated (and a bit taken aback) that he would not come clean when faced with the data.

Thanks, MeFi.
posted by rinosaur to Education (20 answers total) 6 users marked this as a favorite
While it's certainly possible to reset system dates on any number of devices (I've received emails that were supposedly sent years ago, sent from computers with wrong system dates -- ie the computer thought it was 2011 and date-stamped the outgoing email -- looking at the headers confirmed the cause), it seems unlikely that his device's date was reset in between the two assignments in such a way that it looks exactly like he's cheating. It'd be far more likely to set a date in the future, or sometime else in the past.
posted by Mogur at 10:53 AM on February 4, 2016

If the camera's clock is set incorrectly you could certainly wind up with incorrect timestamps, but it does seem pretty fishy after the first confession. I'm not sure what you could do to get a more reliable indicator, either, since now that he's clued in he can set the camera's clock to match the timestamps he submitted, so even if he shows you the camera settings or takes new photos with 2013 timestamps it won't prove anything.
posted by contraption at 10:54 AM on February 4, 2016 [1 favorite]

It could be a wrong camera date stamp. If he wanted to cheat there are so many tool available to edit the metadata.
posted by WizKid at 10:56 AM on February 4, 2016 [3 favorites]

If the datestamp in the camera is wrong, he can prove that by taking a picture of today's newspaper and sending it to you. If the metadata on that shows 2013 then maybe he's telling the truth.

Then again, I'm pretty sure that the dates in photo metadata can be readily changed, so what you might be driving him to do is just learn how to modify the dates on his images before he submits them.
posted by sparklemotion at 11:01 AM on February 4, 2016 [5 favorites]

If he has trouble with the meta-data/year-dates on his camera, this is really his problem to fix. You can only go off of (and grade accordingly from) the meta data that you are provided with.

Ask him to come to your office after class (I assume he brings his SLR to class).
Once in your office, take a photo with his camera.
Look at the meta data on this brand-spanking new image—does it say it was taken in 2013?
And does the date correspond to X number of days since the meta data of the photo that he says he took last weekend?
(like, if the "weekend" photo said it was taken on 01-01-2013, the photo of you would be something like 01-06-2013)
posted by blueberry at 11:03 AM on February 4, 2016 [2 favorites]

Can you perhaps ask them to replicate the photos? The student is either very unlucky (camera date set incorrectly) or not very bright (cheated twice and didn't even bother to adjust the meta data to not get caught the second time around).

Give them a zero - they will learn a lesson either way (check the date on your camera) or if you cheat you will eventually get caught (especially if you are sloppy).

One more question, are you even sure the student took the photos? I would probably do a reverse image search on the web to see if they are just grabbing images from someplace else which fit the assignment....
posted by NoDef at 11:03 AM on February 4, 2016 [2 favorites]

So I guess the issue I'm having is that the dates on the first assignment were 12/29/15 and the dates on the second assignment were 5/23/13. Unless he changed the clock (backwards?) in the mean time ... or somehow it got glitchy and changed itself ... these are old images.

In response to NoDef, I did reverse image search them and no matches came up. He has a pretty robust personal website and instagram, and all his images are consistent in style and editing -- though I suppose he could have pulled them all from one photographer if he wanted to go that route.
posted by rinosaur at 11:08 AM on February 4, 2016 [1 favorite]

Actually, you might be able to catch him out using the ImageNumber EXIF tag, which should correlate to the camera's total lifetime number of shutter actuations when the picture was taken. If you can get a known-recent picture from that camera (ideally make him snap a photo in front of you and then hand you the card, so you know he hasn't mucked with the EXIF directly) and inspect that tag, there should be a sane number of actuations between it and the picture that's ostensibly from last weekend. If the number on the current photo is hundreds of thousands higher, you know he's lying even if he's gotten the timestamp to agree with his story by changing the clock. Likewise, if the number is lower than the one "from last weekend" you'll know he's been messing with the camera in an attempt to cover his tracks.

This all assumes his camera uses that tag, but I think it's fairly standard on SLRs.

Here's a tool that I haven't used but which purports to show full EXIF metadata info:
posted by contraption at 11:08 AM on February 4, 2016 [9 favorites]

This is exactly the sort of situation that college judicial boards are designed for. Every university I've ever attended or worked at has had some sort of procedure to deal with academic dishonesty, and ultimately these procedures are designed so that the instructor ideally does not have to waste tons of time independently investigating these cases. I would compile the evidence you have thus far and just submit it to someone else to deal with so that you can focus your time and mental energy on your students who are actually putting in an honest effort. This will also alert the university in case this student is pulling dishonest behavior in multiple classes (which wouldn't shock me because the level of stupidity/dishonesty here to cheat again after already being caught seems high).
posted by rainbowbrite at 11:21 AM on February 4, 2016 [10 favorites]

Actually, here is a tool that will test just the tag in question. Manufacturers typically make it unresettable by the user so it can be used like a car's odometer, to gauge the amount of wear on a camera for resale purposes. Anything is possible with hacked firmware, though, and of course there's no protection against altering tags on an image file once it's out of the camera.
posted by contraption at 11:35 AM on February 4, 2016 [3 favorites]

Clocks do drift, and if he has a failing backup battery, it is possible for it to drift repeatedly. My pro level Nikon has ended up more than a year off. The ImageNumber tag contraption mentioned should be much more reliable for placing whether the photos are recent or not - any of Adobe's photo products or Picasa will show it.

On the other hand, you're within your rights to just ding him for not checking something that you've already told him was an issue. I was a special snowflake in college so I generally favor a certain amount of coddling of bright but difficult students, but sometimes they also need to learn that poor choices and lack of attention to detail leads to consequences.
posted by Candleman at 11:47 AM on February 4, 2016 [1 favorite]

I don't know about cameras but I know about students. As a former high school teacher, I'd say he cheated again. I'd ask him to my office and say that his story is highly doubtful. He can either confess and go home with a zero, or keep his story and have someone in the university's ethics office/department/whatever pursue this.
posted by CrazyLemonade at 12:20 PM on February 4, 2016 [3 favorites]

Thanks everyone, this has been very helpful. The shutter count idea is brilliant, unfortunately the files he submitted don't have that particular info in the EXIF data.

I especially appreciate the advice that this is probably an investigation for our Academic Honesty Board and not me! Thankfully!
posted by rinosaur at 12:47 PM on February 4, 2016

Any chance he could have previously put the image online, and it's findable through TinEye or Google Images? Their publication dates are from date of earliest upload, and are hard to spoof.
posted by scruss at 12:49 PM on February 4, 2016

Is it possible to play detective and use something in the image that corresponds to 2013 versus 2016? I suspect that in many photos that are outdoors the angle of shadows or foliage of trees or some background detail will give them away. Of course if it is fruit on a table this doesn't work.
posted by dances_with_sneetches at 2:49 PM on February 4, 2016 [1 favorite]

The EXIF data going forward can't be used to ensure the images were recently taken because all of it can be changed with the right editor.

It may be able to be used to trap the student in a lie/inconsistency via the shutter counts and date. And by that the student can't exactly claim to have changed the data to make it appear that they were cheating. Cameras are very good at keeping their data up to date and generally reset to some standard date rather than some random date.

Going forward it's action is going to depend on whether your institute is going to back you up. Place I worked the evidence and history would probably allow you to zero the assingment but nothing more.

If nothing else action at this point even of just another warning will at least motivate the student to cheat better. Changing and examining the EXIF or ITPC data is trivially easy and can be performed with free tools.
posted by Mitheral at 3:26 PM on February 4, 2016

I've been teaching undergrads for ~10 years and in my experience, the really problematic scenarios such as you are describing are surprisingly likely to self-correct.

I've had quite a few cases where I thought a student was probably (possibly?) cheating and worried a lot about the proper response, given my uncertainty, only to find, a few weeks later, the problem solve itself (they flunk an exam, stop attending, drop the class, or similar).

To your original question about metadata being infallible? No: with about 5 minutes and a binary file editor I could fake any meta data anywhere any time. And I've seen enough bugs in software to think "well, maybe?" For example, Excel had a bug for a while where dates in Mac and Windows files would be offset by 4 years:

Bottom line: I would think really hard about your, societal, and the institutional goals - you are probably right and the student has cheated twice. Teach something useful.

If you are in the USA then your semester has just begun most likely, so I would also advise a "see what happens" policy as being reasonable.
posted by soylent00FF00 at 8:23 PM on February 4, 2016

I am curious to know why you looked into the veracity of the student's claims about when he took the photos in the first place. Did something twig you about the photos themselves that prompted you to look at the metadata?
posted by Conrad Cornelius o'Donald o'Dell at 9:28 PM on February 4, 2016

I am curious to know why you looked into the veracity of the student's claims about when he took the photos in the first place.

I wasn't going to check the data in the photos--even after his admission that his first assignment was not shot within the assigned period--except that he submitted his second assignment with a disclaimer about having taken the images before it snowed ... but there was no snow over the assigned period. The lady doth protest too much, methinks.
posted by rinosaur at 5:21 AM on February 5, 2016 [3 favorites]

Hah! That bizarre disclaimer is almost more damning than any metadata.
posted by Conrad Cornelius o'Donald o'Dell at 7:55 PM on February 5, 2016 [1 favorite]

« Older Pick our layover/stopover   |   Should I break up with my cold-footed boyfriend?... Newer »
This thread is closed to new comments.