As you will have realised by now, I am a just little obsessed with investigating the authenticity of images shared on Social Media. Whenever a great controversy arises, I will be there with my magnifying glass, peering into exif data, converting images to text files, running them through reverse image searches and consulting Street View for helpful clues as to whether the image is genuine or not. I admit that it is verging on a compulsive disorder, but so too is the habit of my fellow travellers in forwarding absolutely everything that confirms their worldview without pause.

Most suspect images are not fake in the sense that they are contrived, as in a composite image made up of several individual images or a computer generated image built on a 3D wireframe. Most suspect images have simply been repurposed from their original setting, to lend credence to contemporary claims. Examples include a photo of injuries sustained during a motorboat accident used to prove that riot police ran over innocent protestors with a tank; or a photo of victims of an oil tanker accident passed off as victims of communal violence; or photos of a city marathon presented as mass protests. Both mainstream media and social media users have been guilty of circulating images of this kind, particularly in times of crisis.

Let it be known that in my own investigations, I am frequently proved wrong. Often, after an extensive investigation of a set of images, I have to concede that they are, to the best of my understanding, genuine. In which case I eat humble pie and let them pass me by. That is the nature of the game: I’m not trying to prove a point of view; I’m just trying to get to the truth of the matter.

There is, however, a notable trend in some quarters of claiming that an image has been “photoshopped” when it has not been — to claim that an image is fake, merely because it stands for or against a vested interest. It is rather like that popular characterisation of the science of hadith, which focusses exclusively on the sanad, but completely ignores the matn. In this case an image can be rejected not because it has proved suspect from a technical standpoint, but because it does not confirm the viewer’s preset biases.

To be doubtful about an image is a good thing. But this should only be the starting point. That doubt should force you to investigate, probe and test. As a result of your examinations, you might determine that the image is indeed fake, or you might find that it is real, or you may further conclude that it is impossible to be sure either way. But to claim and pass judgement with absolute certainty in the space of a few seconds that an image has been “photoshopped” simply because you wish to believe one way or the other is an invalid measure.

I am a dab hand with Photoshop myself, although my skills are somewhat “old school”. Most of the techniques I use have been superseded nowadays by clever algorithms that do the work for you, or individual apps that can clone and heal at the touch of a button. Nevertheless, my intimacy with the actual pixels of an image, rather than an array of clever filters, has trained my eyes. I can tell when there really is something wrong with an image, or whether it is an optical illusion. Many a time people claim that an image has been “clearly photoshopped” when really it is very hard to tell.

Such a case has occurred recently when images purporting to be screenshots of smartphone conversations between a popular preacher and a number of women were published on an anonymous website set up exclusively for the purpose of exposing him. Immediately responses divided into two major camps: his supporters and followers, who claimed that the images were “obviously fake”, created by his enemies to discredit him and his work; and his detractors, who were convinced that the screenshots were real, and needed no further evidence before proclaiming his guilt far and wide.

In reality, neither camp had a firm foothold on the truth of the matter, for the images in question prove nothing on their own. They could be real or they could be fake, and there are means to achieve both outcomes.

On Facebook, one individual went to great lengths explaining why he believed that the images had been crudely created in Photoshop, citing problems with colour, spacing and alignment issues, and the use of the wrong fonts. Though I initially wondered about the fading colours of speech bubbles in the iMessage screenshots — as a non-user it looks odd, but it turns out to be accurate — I personally believe that his objections are unwarranted. There is in fact no real evidence to support the idea they were faked in Photoshop.

One way to learn more about a file is to rename it as a text file and then open it in a text editor. Sometimes, if a file has been exported for the web after manipulation in an image editor, you can obtain all kinds of useful meta information about how it was created. Here, a general-purpose language for representing information on the internet, called the Resource Description Framework (RDF), comes into play.

In this case, most of the purported screenshots do not yield this data. You will instead encounter several hundred lines of unicode gobbledygook interspersed with Apple logos and mathematical symbols, interesting only to the conspiratorial steganographer. However a couple of the images do reveal RDF data, allowing me to learn the following about a particular file:

  • The image was edited in Adobe Photoshop CC 2015 on a Mac.
  • The file was first created on 9 September 2017 at nine minutes past midnight, US Eastern Standard Time.
  • The file was modified just after an hour later at 1:21am, US Eastern Standard Time.

Now those in the supporters camp might say that this is an open and shut case proving that the image has been faked in Photoshop, but that is not necessarily the case. All it proves is that the image has been edited in the application during its lifecycle — and an innocent explanation can easily be found: in this particular screenshot, names have been blocked out with a black line tool. Having annotated the image in this way, it may then have been reexported as a flat PNG file via the “Save for Web & Devices” menu, at which stage the application may have added the RDF data to the file.

The other screenshots, meanwhile, may have been annotated and saved natively, without exporting for the web. This could also explain the difference in quality between the various screenshots: on close inspection, some of the images are very lossy, with too much noise around the text in the iMessage speech bubbles. To some observers, that is the golden bullet and proof of mischief at play. But again, this does not prove that the image has been faked in Photoshop: it just as easily means that when the file was saved for the web, a low resolution PNG setting was chosen, which significantly downsampled the original image.

So does this all mean that the screenshots are genuine and are what they claim to be? Does this mean that the detractors’ camp is right? I have to say, absolutely not.

For in reality, the claim that the screenshots have been faked in Photoshop is a bit of a red herring. The fact of the matter is that if you wished to fake a WhatsApp or iMessage conversation, you wouldn’t use Photoshop at all, for there are numerous apps available for iPhone and Android which enable jokers, hoaxers and the malicious to create fake screenshots with ease.

The very best of these apps will create conversations that look identical to the real thing. A data forensics team might be able to determine the difference, but there is nothing most of us can do to determine the truth. A screenshot of a fake conversation is still a screenshot, after all, containing the same data encoding as a screenshot of a real conversation. The only avenue left to determine the truth is to rely on the honesty of those who generated the conversations in the first place, which is rather difficult if everyone involved claims anonymity.

So can we say whether these screenshots are real or fake? Not really. For those in the supporters’ camp, the content of the alleged conversations is evidence enough that the images are fake. For them, the matn is clearly spurious and contrived. Which is fine, I suppose, if the man’s reputation precedes him. For those in the detractors’ camp, meanwhile: well the matn is absolutely convincing and the screenshots absolutely real. But neither position establishes authenticity here; one’s truth depends purely on the emotional relationship of the observer with the subject.

There are few facts accessible to the detached observer upon which we can agree. A website was set up in the name of the accused, purely for the purpose of exposing him. Muslims of certain sectarian persuasions dedicate their lives to exposing people for heresy and misdemeanour, so this is not so unusual in itself. From the Whois data, we know that the domain name was registered with a US-based registrar, while the website itself was hosted by a UK-based web host. Beyond that, for the stranger, everything else is speculation.

Particular attributes might lead an investigator towards one position or the other. If the images are fake, then whoever created them certainly went to a lot of trouble, using two different versions of iOS, two different messaging apps and three different network providers: one in Britain, one in India and one in the US.

On the other hand, I suspect that if somebody was seriously intent on maliciously smearing another’s reputation, they would go to whatever lengths they deemed necessary and take the time to do it properly. Though of course this is not always the case: prosecutors for Turkey’s Ergenekon trials a decade ago produced evidence that had been faked so shoddily that in the end sentences had to be quashed, although only when it became politically expedient to do so.

At the end of the day, the objective observer can adopt neither position or back any particular side. The evidence, as it stands, is not irrefutable. It may be convincing or plausible, or conversely, it may completely suspect and improbable. Witch hunts do happen; people who find themselves on the wrong side of a sectarian or political divide are frequently criticised and taken to task; and proponents of an unpopular science often face unparalleled attacks. But so too do all of us have the capability to become corrupt, to lie and cheat, to let power and ego go to our heads, to sin and err and make mistakes, for we are but human.

The purpose of this article is not to declare the facts either way. About this particular individual, I know very little, other than what I have gleaned in passing. It is not my place to be drawn on what really happened: that is between him, his mediators and those directly involved. His case has been thrust into the public eye by comments he published on his Facebook page in response to comments published elsewhere, compounded by a website set up to expose him. That chain of events in unfortunate, making public what need not have been.

Nevertheless, I hope to shed light on the difficulties inherent in sifting fact from fiction in the age of digital media. Sometimes, using a variety of freely available tools, its is easy to trace the origin of an image. The history of a photograph can often be traced backwards with a reverse image search. A composite image can usually be identified under careful examination. A poorly edited screenshot generally stands out like a sore thumb.

But it is not always so. Sometimes it is nearly impossible to determine the veracity of an image. CGI is so good nowadays that to the untrained eye attempting to substantiate the facts is almost futile. Even twenty years ago, a person might have been convinced that Forrest Gump met Richard Nixon, but today long-dead Hollywood actors are regularly brought back to life to complete a film. Nowadays a screenshot of a conversation can easily be faked, and nobody would ever know.

But of course, just because they can be, it doesn’t mean they are. The claim that an image is fake is just as much as a problem as the claim that it is true. Some of the tools we have at our disposal can be used to show that something claimed to be fake is in fact true.

Where claims cannot be verified, however, I would discourage people from adopting a hard definitive opinion in either case: if it is impossible to ascertain the truth in a matter, do not embrace an implacable stance based purely on emotion and attachment. Try to be just and reserve judgment; admit that you do not know. Avoid consolidating entrenched opinions.

In a case such as this one, we might easily be persuaded to leave alone what does not concern us: to mind our own business and move on. But in other cases, the effects could be more far reaching. What if an image or a screenshot of a conversation had a bearing on war or peace? What if it related to a person’s or a community’s safety? What if a fake screenshot led somebody to cause another person harm? Or if a genuine screenshot, claimed to be fake, resulted in a vulnerable victim being ignored.

These are the very real issues which confront us in the age of digital sharing. An entrenched opinion, formed by emotion instead of evidence, could easily be our downfall. So when you do not know, know that there is no harm in admitting it. It is better to admit to ignorance, than to insist on a truth which may not be so. And it is best to realise that “it has been photoshopped” is a lazy defence, when all that has really been air brushed are our biases.