About Teletortoise

Teletortoise is an art installation designed to monitor how participants remotely interact with a testudine through the manipulation of its environment.

The piece contains two distinct experiences. At its most approachable, Teletortoise aims to rekindle some of the magic of the exploratory internet, a virtual world of new discoveries and novel interactions. However, through instructions designed to disquiet the audience, the piece also inquires into many aspects of how human’s experience the internet.

The artistic goals of Teletortoise were to excite web users; to explore human reactions to being given control over a creature’s life via the internet; and to ask questions concerning typical internet experiences.

The technical goal of the Teletortoise project was to successfully create a session-based, web-distributed piece of robotics that allowed data tracking.

The Teletortoise experience explained

A webcam stream of a turtle enclosure was made available to all viewers of between 6pm and 7pm, from May 1st – 7th. The website introduction the project and warned participants that the creature (advertised as a water-adverse tortoise) was in life-threatening danger should they allow the water-level to get too high.

While the work was live, audience members were randomly selected to “direct” interactions with the turtle for a five minute period. For their allocated session, directors were allowed to toggle light, food and the flow of water into the enclosure between two states: on and off.

When toggling the water on – or when making another selection while the water was still flowing – participants were reminded that leaving the water on could harm or kill the tortoise. The warning also prompted the participants to turn off the water flow.

When their session expired, all variables reverted to their default “off” state, although the water level inside the enclosure did not drain out. This meant that the water level increased throughout the hour period.

All interactions were recorded, including the choices of users, to allow analysis of how people chose to interact. The analysis of this data, and feedback from participants, formed the second part of the work.

How Teletortoise was made

The technology behind Teletortoise successfully created the required experience to convey the purposes of the piece. It used six programming/coding languages: PHP, HTML, CSS, Javascript, Processing-assisted Java and Objective C for the Arduino, as well as electronics knowledge to assemble the robotic elements.

  • PHP (running in Apache on a local PC) successfully stored user-submitted data as variables on the local system for analysis and control of the enclosure
  • Javascript successfully prevented multiple actions from being submitted at once and improved the experience of participants
  • Processing successfully read the PHP-saved variables and transmitted the information via Serial to an Arduino
  • Objective C on the Arduino successfully received commands over Serial and called different functions to either turn on or off the light, food and water flow
  • The robotic elements – an LED, a servo and a solenoid valve – functioned as designed
  • The web stream was live without error throughout the sessions

There were numerous issues both before and during the experiment, however:

  • Video streaming lag was higher than expected. Depending on the provider, it ranged from between 15 seconds to one minute. I opted for the best possible provider, and compensated for the lag by adding an 18-second wait after participant action submissions. This prevented participants from issuing a second command before the video could display the first (actions were transmitted from the web to the device instantly.) While the lag was disappointing, it also added a new dimension to the experience. Water-based interactions were more stressful, as every time the water was turned on it was had to run for at least 17 seconds.
  • Processing proved only semi-reliable. It crashed three times during the installation, preventing participant commands from being pushed to the Arduino. By monitoring the Processing console, however, these were always remedied within three minutes. On a positive note, when one crash caused the water to fill the tank for four consecutive minutes, a participant reported a strong emotional reaction, contacting me to stress their concern for the “tortoise” and reporting feeling “stressed and anxious”
  • Parts of the website were incompatible with iPhones

You can download the source code here: It’s designed to work in the root of a web directory, so if you want to put it in another folder you’ll have to amend the locations. Otherwise, it should be good to go to save three button toggles as booleans (1 and 0), and have the Processing and Arduino code respond.

Additional comments on Teletortoise

Teletortoise successfully caught the imagination of internet users. In just seven days, with no prior promotion, the project generated over 1,000 unique visitors, over 130 shares (Facebook and Twitter) and over 80 different interaction sessions generating nearly 1,000 interactions. It also generated discussion on Reddit and inside chat rooms. This means it successfully completed one of its artistic objectives – to excite and engage visitors.

Teletortoise’s conceptual questions, and their degree of success, are outlined below. It should be noted that participation was deliberately anonymous in order to allow the participants to behave without an external censor. Therefore it’s difficult to gauge how the work was received and how well these questions were received. Also, due to the nature of the work, any understanding of the artistic intention prior to participation would have biased the interactions, and therefore there are questions over the artistic merit of the concept for much of the audience.

The following responses were gathered from people who elected to discuss the project with me after it had finished, and through Facebook and Reddit comments and internet chat rooms, and are offered only as anecdotal evidence.

How real is what we experience on the internet real? Does it matter?

Information on the internet is unexperienceable, and therefore the truth is subject to abuse. Teletortoise explores this by existing between the realms of reality and fiction: the interactions are real, but the advertised danger is false. All visitors have to choose which elements of Teletortoise’s truth they subscribed to.

Feedback from participants could typically be organised into four groups:

  • Users who believed the website’s text, and believed the animal would be harmed by water
  • Users who didn’t believe the website’s text, but believed that the animal would be harmed by water
  • Users who didn’t believe the website’s text, and didn’t believe the animal would be harmed
  • Users who didn’t care about either the website or the animal

Disbelief was gently encouraged in the piece, as all images on the website and in the promotion material, and the animal itself, were water-dwelling turtles, and not of water-adverse tortoises.

I believe this question was successfully experienced by participants in the second and third categories, who typically went through a cognitive reasoning process when choosing to disbelieve the website’s text. Some people in these categories also:

Disregarded their own knowledge, identifying the animal as a water-dwelling animal but still be concern regarding adding water into the tank

Questioned whether the creator of the project would let this happen (see below)

Where does a web user’s power end? How can an end-user have power on the web?

The obscure situation of a tortoise’s life being under threat from internet controllers created a typical response from participants: would the project’s creator allow remote users to drown his tortoise?

I believe this successfully encouraged participants to ponder how much power their interactions would grant them. There was no way for a user to know whether the tortoise would be allowed to come to harm (and thus their interactions were unregulated), or whether they were secretly being limited.

The only way for participants to truly know how much power they had – and whether the creator or his mechanisms would step in – would be to flood the tank, threatening the tortoise’s health. Is it only by acts of barbarism – of conducting offensive acts – that web user can truly feel empowered?

I believe this question was also successfully conveyed; not only was it stated by web users, but many users elected to flood the tank, and some participants went as far as to formulate theories as to how drowning would be avoided, imagining sensors, drains and fail-safe shut-off valves.

What responsibility comes with internet-based interaction?

While most of the participants questioned were more concerned about not hurting the animal, I believe the piece was well-designed to encourage questions about responsibility of actions on the internet. If the animal was hurt, who would be responsible?

The creator put the animal in this position and made the actions possible, but it was the visitors to the website who collectively raised the water level despite warnings on every occasion the work was online. As Marina Abramovi? described of her Rhythm 0 work, “What I learned was that… if you leave it up to the audience, they can kill you.”

Some might rationalise that Abramovi?, as a sentient being, put herself in a dangerous position and therefore is obliged to face the consequences. With an interaction with an animal, however, there is no such excuse.

What is responsibility on the internet, when there are many places where blame can be attributed? Does blame shared mean blamed halved, and with distributed systems, does that mean we can never accrue enough blame to appreciate the consequences of our actions, to feel guilt?

Although participants’ level of responsibility wasn’t clear, it did successfully generate emotive responses from visitors. These were in the form of questioning the ethics, declaring the piece “hard to watch” and even reporting feelings of stress because of the consequences of their actions.


While I believe that it was conceptually a success, I was disappointed with the aesthetic of the final video. Video streaming limitations (as discussed below), along with difficult lighting conditions (more light was needed to compensate for the video quality, causing reflections in the enclosure) meant that the video stream was less appealing than I would have liked. However, the low quality of the broadcast may have helped participants believe in the reality of the piece, as if the safety of the animal had not been considered within the amateur setting.

Technical limitations (discussed below) also forced a delay in the video showing the commands, which also broke some of the immersion.

I was unable to reconcile a way of deliberating engaging visitors was the conceptual aspects of the piece. Explaining the artistic purpose would surely have biased interactions, while collecting their information would have removed the anonymity that I also deemed important for free expression on the website. However, I believe many visitors experienced and understood conceptual aspects of the work, such as the raw emotional response.


Teletortoise - by Jack Ratcliffe


Special thanks go to sinmente0987 and lolamadeus for their sound effects, and Freepik for the icons.