>> Don’t test the blue bird

This project as a whole is named “This Was The Way Their World Ended” – a play on Halo’s iconic quote from T.S. Eliot’s The Hollow Men:
“This is the way the world ends: Not with a bang, but a whimper.”

What took place December 2008 through August 2009 wasn’t what was initially designed, and it consistently grew and evolved through its life span, enduring many growing pains. To begin, we’ll look at some of the most significant difficulties and failures from Intimation’s execution and what we learned from them. First, our utilization of the blue bird: Twitter.

The community who followed at the Unfiction forums decided to refer to the project as “Intimation”, coined after the primary point of contact: a lonely little Twitter account named @intimation.


As a social networking tool, Twitter also has a fountain of storytelling potential. For ARGs, it’s become a significant tool for news, interaction, puzzles, revelations, coded messages, and more.

It’s a very beneficial resource because it’s a pull-based personalized platform, as well as a simple broadcasting tool. It’s no surprise therefore that fictional Twitter personnas are popular. With so many ARG players now also actively using Twitter, it would seem a natural and believable thing to have an ARG character adopt a Twitter account.

But while it sounds natural to have a character active on Twitter, there’s still a general etiquette to follow, and if they don’t abide by it, the little blue bird can turn into an angry little blue bird.

The Launch: @intimation pokes around

Our launch was a major oversight that left a sour first impression.

As a throw-back to I Love Bees, our intent was to have an AI ‘crash land’ in our time and take up residence… somewhere. This AI’s purpose was to be primarily informational: share updates, tell a bit of a story, and lead up to the launch of the ARG as a prelude. We had the AI take up residence in our 2008 internet, working to gather itself and carry out its mission.

Unlike ILB however, this AI didn’t land in a website — it reached out to embrace a communication method that would reflect its tendril-like network functionality; similar to Melissa’s introduction in ILB, but intentionally different. That tool was Twitter, of course! What better way to gain attention (rather, to launch a rabbit hole), than by an AI socially ‘poking’ random people as it “feels its way” around our established networks.

Following in the footsteps of the “Smart AI” properties introduced with Melissa, our AI was somewhat disoriented after its crash landing. It also turned out to be very inexperienced in the ways of Twitter’s social etiquette. It dumped (aka tweeted) some cryptic phrases, and then followed a number of Twitter accounts.

A large number of Twitter accounts.
Halo people, ARG people, story people, industry people.
And then unfollowed them.

It made sense in character – feeling and poking around, testing the waters, “wetting the system” as it were. In our mind, hoping not to imply connection with any particular individuals, immediately unfollowing accounts removed those possible implication which could be gleaned through the list of who it was following (something sure to be checked by curious ARGers).

A few Twitter users reciprocated, following @intimation in return, but most didn’t. The first round wasn’t very effective, and so the AI tried again, following many of those same people a second time. This time a few more followed, but instead some of the key community folk who typically take note of strange, mysterious things and potential puzzles, rather than becoming interested, became quite annoyed.

While we watched for mentions of @intimation in blogs and forums, its first mention at the Unfiction forums was not too favorable. The launch was seen as confusing and annoying. Standard Twitter conventions were broken. Even though the AI’s actions may have been in character, they were themselves perceived negatively in the context of that platform, and turned many people off right from the start.

Even the launch puzzle itself was, for the purpose of an introductory puzzle and mystery, much more difficult and vague than it should have been.

Hollow roadblock, cryptic dream

The puzzle came in the form of a dream-like data dump. It included a series of lines from The Hollow Men, containing a plethora of typos. These lines were tweeted one at a time on a schedule, but there was no way to solve it until the complete puzzle had been released. It was merely an odd text dump growing lengthier, tweet by tweet. The hopeful intent of this action was to spark some discussion and speculation among the curious, until the final component needed to solve the puzzle was revealed. Unfortunately, even after that point it proved just too obscure a puzzle to piece together, let alone easily and quickly.

In other ARGs, puppetmasters have easily become frustrated at stalled situations, even giving the impression of blaming players for not “seeing the obvious” to move forward. We didn’t want to fall into that trap ourselves because of puzzle design. We also didn’t want to break the 4th wall, or do something out of character and even more confusing, pulling our players out of the experience.

So, we adopted another property of the AI demonstrated with Melissa in ILB – we introduced a more advanced form of the System Peril Distributed Reflex.
This character quickly became an asset for interactivity, a fallback option to provide responses to players and serve as a guide in the case of roadblocks. But in-story this ‘maintenance process’ was activated to help ‘fix’ our AI. This emergency system process nicknamed “Sys” effectively took control of this situation.

In a sort of desperation, Sys reached out to Youtube, and provided a rendered dream sequence, a visualization of the cryptic data dump which it labeled “foreboding dream”. This video re-iterated in its description all of the poem’s lines (typos and all) so that there would be no confusion (we hoped) and to serve as a single source for people to examine and further attempt to solve the puzzle.

But time passed, and with no solution in sight even a month after the tweets began, nor movement in any other aspect of the story, the existing community started to grow tired of the mystery.

So, we then had Sys render yet another dream sequence, as if the AI were being haunted by this unsolved mystery (which was not far from the truth), and that interpreting (solving) it was essential to Sys’s maintenance objective. Thankfully, after this hint was provided, the puzzlers had the needed breakthrough and the foreboding dream was interpreted, ultimately leading to the website and Sys’s somewhat meta archival hub, thewaytheirworldended.com.

We’d hoped to attract a larger audience at launch, but had failed to keep it easy to access, straight forward, interesting, and welcoming for that scale. We created too much of a challenge in an insufficiently designed puzzle for an audience that wasn’t yet sufficiently invested, and so risked losing their attention to frustration or boredom. But we pushed through and quickly moved on in an effort to regain any lost ground.

It was a rough start, and having to employ a level of damage control right from launch set what in hindsight felt like the precedent for future development and execution. However, our damage control also had the side-effect of providing an opportunity to expand the story and its execution – although this strategy become a dangerously recurring one.

As it stood, we’d now established Twitter as our central interaction platform, yet the drama was just getting started.

Private Uplink: Communication Failure

Privately communicating among the puppetmaster team via Twitter direct messages using SMS text messaging is, simply put, a risky move.  The following important lesson was learned the hard way in July of 2009 around the climax of the ARG, in Portland, Oregon.

One must never forget that when using Twitter via text messaging, you must begin private messages with “D [username]”.

To this day I don’t know how many people noticed, though I’m sure someone must have, that during ARGFest 2009 in Portland, a rather public and revealing tweet, sharing details about a certain PM, the Intimation ARG and a certain event involving hidden skulls (see Part 9: Hunting skulls), was posted publicly for all to see.

This sensitive message was deleted within seconds once realizing the error, however at that time even if a tweet was deleted within seconds of its posting, it would remain in the public index of search.twitter.com for 24 hours before being completely removed. Some Twitter clients used that engine to pull tweet timelines and search results, so this tweet was still readily available on a simple search. And so, for 24 hours during ARGFest the tweet sat, I deathly nervous, hoping no one would take notice of it. Thankfully however, the planned event and scavenger hunt pulled off without a hitch.

Lesson learned! It’s certainly safer to avoid using standard text messaging to send sensitive private Direct Messages through Twitter.  Today we have capable smartphones and apps which provide that ability much more effectively and efficiently. Even so, we still see it happen – replies to private conversations are occasionally posted publicly to Twitter by people mistakenly. Occasionally, entertaining. Potentially, devastating.

Etiquette: Broadcast Overload

Lesson #2: Twitter is not an instant messenger! That, at least, was the generally accepted etiquette. For users with common friends, it’s usually considered an annoyance to publicly carry out lengthy multi-participant conversations, even though Twitter is certainly capable of being used in such a manner.

Public tweets and private messages between two accounts are Twitter’s two methods of communication. When using Twitter for character interaction, excessive public tweets could quickly be considered as spamming if followers become uninterested or annoyed at a stream of conversational tweets they are not a part of in their timeline. If you decide to carry out public conversations with someone, you risk annoying common friends and others who aren’t included in the conversation. On the other hand private messages, while personal and more intimate, are inherently limited in reach and visibility – you lose the exposure you may want by using Twitter in the first place.

There’s really no happy middle for active and public, yet directed Twitter communication, without risk of annoying bystanders. It’s safer not to try to repurpose a platform for a means of expression that conflicts with accepted social etiquette. If you do, you risk losing people’s attentions, either from confusion, or pure annoyance.

In our case, Twitter was the primary means of communication with the AI. At the very least, this platform for communication shouldn’t be exclusive and forced on people. If you give people the choice to listen, you also give them the choice to ignore, even boycott. You risk losing their attention, or that of their friends, or their friends’ friends. This really applies to most any social platform. As tools, they can produce favorable social and viral results, or equally negative consequences if not used responsibly.

As for exclusivity, not everyone owns a Twitter account, and there are still many who prefer not to. We’d planned the prologue communication to be carried out entirely on Twitter, but we noticed a number of people who were interested in the story and wanted to enjoy it but didn’t like or use Twitter. They – by choice, even – were excluded from a heavy interactive element, left only to rely on second hand reports from other players.

Ultimately, we came to realize that having our character Sys communicate through Twitter worked best when its messages were publicly intended for everyone, to the point, and broadcast on a periodic, extended schedule people could expect, rather than used actively for public player interaction that spilled over as undesired updates into their friends’ feeds.

To address the issue of having non-Twitter players, we also had Sys’s public tweets duplicated at its hub website in an archived state for players to easily access without using Twitter. For individual communication, we resorted to private direct messages, and hoped that anything important we revealed would be shared with the community.

Final Twords

Back in 2008, the manner by which we implemented the Twitter platform for ARG storytelling we found to be a risky means for communicating and interacting with players.  Twitter is a powerful broadcast medium, but a touchy medium for open group interactivity, and we made a number of etiquette mistakes that hindered acceptance and entry to our project.

Even though Twitter has grown and evolved since then, some things still remain the same. If being implemented in a storytelling campaign, whether interactive or expository, it should integrate with the players’ own personal use as seamlessly as possible. Pick a communication style – broadcast or interactive, automated or personally maintained – then avoid making any dramatic changes. Otherwise, it can stick out and risk being thrown away and ignored as an annoyance, not only by the individual, but also by those to whom that individual has reach. Twitter is a unique beast, combining both strengths and weaknesses of a typical blog medium with those of a chatroom. It’s a broadcast platform, but it’s also pull-based, personal, and interactive.

Respect the blue bird.


PITFALLS: Transparency, and Bungie.net