For several weeks now I’ve been (half) joking on twitter that I’d like to have a way to have my network data objects decay and fade, if not distort and become unreliable. I’ve also talked about the idea of fake data trails, where a network of bots creates false locations, friends, transactions, in order to mislead those who would use my information for their own ends. I’ll return to the second idea in another post, but here I’ll flesh out the first one. This post sketches out the idea of intentional unreliability.

So the first question is this: when everyone is continuously looking for ways to secure their data, to make it more reliable, authoritative, to reduce noise & spam, why would I want to introduce less reliability and additional noise? Because I want to be the one that determines how reliable or unreliable my information is. And because much of the information about us out there festering in databases is unreliable - I want to manage my unreliability. (And yes I understand this is mostly impossible, like controlling the weather, but read on).

We are used to the idea that physical media objects like books and CD’s are not mutable over time. I can’t change a word or a note of something that has been published in the analog realm. Some of this expectation carries over unexamined into our relationship with networked digital media objects that we create.  We are still adapting to the idea of dynamic media objects, ones that change over time, either by being updated with new information, multiple authorship, or other means. What I want to be able to do is to create objects that have a life to them, with parameters I can set either by myself or in concert with others.

Data Effects

A loose analogy is one of adding audio effects to an audio track, a vocal track for example. You can add equalization, reverb, pitch correction, and so on. The source sound is non destructively altered and the effects can be automated and changed as needed. Similarly, for data, I’d like to be able to add controls that allow my information to change over time. I want bugs in my own information.

These data effects would allow information to decay, to be fallible, like memory. It would allow  some control over how data lives in the network over time, frustrating efforts of those who would use it for commercial purposes, for example. So Instead of my information becoming some else’s market intelligence, it becomes market disinformation. I become: just not worth it.

Expressive Parameters

So if there were an app for this, following our audio analogy, the app would have faders like on a mixing board. The faders would set parameters for pieces of data, like a photo, pdf, tweet, or blog post and control parameters like:

Decay  -  amount of deterioration of the object over time. So an image would start to have random noise appear in it, a text file might have random characters and words appear (for example: 10% over 6 months, or 100% over 2 years, etc.).

Unreliability / Distortion  -  % chance that the object will fail to load or appear, or that parts of it will be damaged or gone altogether. (for example: Object x has a 4.4% chance of failing to load/appear and 18% chance that 6% of it will be damaged, etc.).

Unpredictability  -  An idea borrowed loosely from particle physics: % chance that the object will show up somewhere else, on a different server / service / site, or that its specifications will change. . It may do things like blow up in size, from 4 MB to 2 GB. If an image, it may scale way up or way down, or show an entirely the image at random.

Fade  -  this is more for images and video, how long it twill take for the image to fade to  a certain % of white or black. It could even fade to another image with a statement, logo, link, etc. It could fade to a default image that said something like “this image has been forgotten”, for example. Or it could result in a data shroud, with a slight impression of the former image.


There is no application, of course, and this is mostly a way of making a point via absurdity. That said, If I had my way, something like this would be built deep into various standards, platforms, and filetypes. None of this will ever happen but it gets me thinking about a potential class of objects that have a kind of remote control, or generative properties that we set, either as individuals or in groups.

I don’t just want technical control, I want expressive control over my information. My unreliability will have style.