Masters degrees, card carrying, secret-handshake knowing designer here. I remember when the notion of dark patterns was first discussed and differentiated from anti-patterns and I think this current discussion is quite ironic.

It's ironic because it is hinging on a narrowly constrained definition of "user interface" design - one that has been appropriated by web/app designers in recent years. UI goes well beyond what is IN the browser window. Discussion of user interfaces occurred before they were graphical user interfaces, well before they were web browsers. So for a discussion that hinged on the idea of the misappropriation of terms, this scores high for irony, IMHO.

SO, now that we're clear on the idea UI's can include many things, it's also clear that the URL and the items return based on the url requested absolutely fall under the notion of UI. And, this is most definitely an example of UI manipulation to create unexpected and negative results for the user to the benefit of the website.

Additionally I should point out, the whole notion of interaction design once meant something much broader than digital interfaces - as an easy example, look at Don Norman's early work and you'll see that interaction and interfaces go well beyond windowed interfaces or even digital interfaces.

The Psychology of Everyday Things (later retitled "The Design of Everyday Things") is indeed a great book. I call out the original title, first, to be a hipster but more to point out that his very aim was to talk about things not Human Computer Interaction. It is not a book primarily about Human Computer Interaction and your regarding it as such or at least lumping it in with HCI shows a serious misunderstanding of the larger point.

It isn't that you can think of many things as interfaces it is that many things are interfaces and interfaces were around long before they became graphical or even digital/computer-based. So, URLs themselves are definitely interfaces – they're UI's and machine interfaces as well, given their multiple roles.

For the concept of Dark Patterns, the manipulation you're talking about is, at its core, abusing convention, expectation and perception to steer people into an experience that they wouldn't choose if it were more obvious. So, essentially, dark patterns are deceptive.

What I gather you're asserting is that they must also be manipulative and get people to do something, themselves? By that bar, I think the imgur returning of pages instead of images when the convention is to return an image for a url ending in .jpg, etc. may be questionable. No URL request will ever rise to the level of nuance that a visual interface will. However, I still think this case fits. It is abusing a set of conventions and intentionally guiding a user into something they weren't expecting.

Additionally, think about the multiple use cases here. It's easy to focus on the casual browser clicking a link from Reddit but it's also about the user creating the reddit post. They are following convention, using what they think is an image link, choosing to post it, only to then unwittingly be involved in serving up that annoying as hell moving cat paw ad on top of the image they're trying to share with others. That sounds a little like a dark pattern at work to me...

> It is not a book primarily about Human Computer Interaction and your regarding it as such or at least lumping it in with HCI shows a serious misunderstanding of the larger point.

The primarily I did not say, did I? The book was required reading at the first university I heard a HCI lecture. It was recommended at my masters (in HCI, both the masters and the lecture), and contents from it were teached. I'm not sure about everyone in my current team, but I know at least some have read it, and I saw general references to its content. And those guys are pretty much the core of european academic HCI.

Everyday things has that role, as I could see, exactly because it is not talking about computers. It manages to show concepts and principals in a way that makes clear how they are universal. Besides, computer interaction does not happen only through display interfaces.

> I call out the original title, first, to be a hipster but more to point out that his very aim was to talk about things not Human Computer Interaction

Actually, the foreword of the 2002 edition I have open right now explains the change of title. It describes that it is because psychology in the title made it go to the wrong bookshelf in stores, in that people did not capture that it was talking about objects and design instead of psychology itself. Where do you have your explanation from?

> What I gather you're asserting is that they must also be manipulative and get people to do something, themselves?

Exactly. Or to not do something they'd usually do.

> That sounds a little like a dark pattern at work to me...

A little, yes. I stand by that manipulation is not enough included, and deception is not enough, and the link look is not enough. But like I said, I now understand why others are mixing it up – and the point with the different use cases might apply. I was serious in my explanation in another comment that there might be a group of people for who it really works as a manipulation, even though I can't see it working in a general way without prior conditioning, and then it fits the dark pattern definition reasonably well.

I think that was his point, different people can define the same word in multiple ways. Just because you/your favourite source says one thing, doesn't mean everyone else agrees, for better or worse.

Actually my definition of "definition" is that it is strict.