Structured Listening

Three numbers, one climb

Sodapoppin, November 2018

• 27,934 concurrent viewers, then 28,076, then 28,203
• Two hundred sixty-nine viewers in a hundred and twenty-two seconds
• The dataset records the climb to the second
• What the dataset cannot tell you is what the climb was

Same rows, four different events

• A raid: another streamer ending a broadcast and sending their audience over
• A clip going viral on Reddit, pulling in strangers
• The streamer returning from a break
• A game update, or just the slow accumulation of a good night
• The row looks identical in all four cases: a number, a larger number, a larger number still

Signatures you can only learn by watching

• A raid has a signature: a near-vertical jump, often with a wave of near-identical greetings in chat
• A viral clip has a different one: a slower swell of viewers who do not know the room’s conventions
• The dataset will not label these for you
• You bring the labels, and you can only bring them if you have done the looking

Immersion and structured listening

• Immersion: sustained, systematic attention to a subject before analysis begins
• For a content analyst working with Twitch, immersion takes the form of structured listening
• Watch streams, read chat, take notes, with the discipline an ethnographer brings to a field site
• The dataset is historical, 5.5 days in 2018. The platform is live in front of you right now.
• The behaviors it records are still there to be watched

What happens when you skip it

A coder who has never watched Twitch reads the log as plain text and:

• Sees KEKW and cannot tell an emote from a typo or a username
• Sees thirty accounts posting the same phrase and does not recognize copypasta
• Sees “@sodapoppin same” and cannot tell agreement from a reflexive stock reply
• The scheme sorts every message confidently, and the buckets do not mean what you think

Thick description

• Geertz (1973): a record rich enough to capture not just what happened but what it meant in context
• A thin description of the chat log counts messages
• A thick description knows the count is made of raids, inside jokes, and reflexive replies
• Structured listening is how a thin dataset becomes something you can describe thickly
• The medium changes. The discipline does not.

Mode one: casual watching

• Initial exposure, nothing more
• Open several streams across the range: a large gaming channel, a small one, Just Chatting, art or music
• Do not take notes. Do not try to code anything.
• Let the medium be unfamiliar. What surprises you? What convention do you not understand yet?
• The only goal is replacing your assumptions about Twitch with exposure to it

Mode two: focused watching

Choose a few streams, watch with one dimension in mind per pass:

• Chat behavior: how fast does it move, who is it addressed to, when does the pace change?
• Viewer trajectories: does the count climb, hold, or fall, and what is happening on stream when it moves?
• Host moves: does the streamer read chat aloud, react to it, ignore it, steer it?

Brief notes after each pass. You are still describing, not coding.

Mode three: analytical watching

• Watch a larger number of streams, asking what recurs
• Do certain chat behaviors cluster with certain kinds of stream?
• Are there category-specific conventions, so competitive-game chat differs from art-stream chat?
• Which cases resist easy description?
• The contours of your coding scheme start to appear here

Field notes exist or they do not

• Observation without a record is just watching television
• Field notes: written, dated, specific accounts of what you saw and what it made you think
• They are not polished prose. They are thinking on paper.
• Their value is catching an observation while it is fresh, before it hardens into something you assume you always knew
• The notes do not need to be good. They need to exist.

Observational and methodological notes

• Observational note: what you saw. “Watched a mid-size Just Chatting stream for forty minutes. Chat moved in bursts, near silence while the streamer talked, then a flood every time they paused.”
• Methodological note: a measurement problem. “The bursts after a question are answers to the streamer, so directed, but also performance for the room. A coder reading the log as text cannot see the question. The codebook may need a rule.”

Theoretical and comparative notes

• Theoretical note: links an observation to a framework. “The burst-after-a-question pattern fits uses and gratifications (Katz, Blumler, & Gurevitch, 1973). Chat volume is not a steady trait of a stream but a response to host moves.”
• Comparative note: how cases differ. “Two streams, same hour. Gaming chat ran fast and reactive; art chat ran slow and conversational. Gaming and non-gaming chat are not the same object.”

The rhythm

• One plain-text document, dated, one entry per observation session
• The V2V Hub provides a field-notes template with the four note types
• Set a rhythm: a note every fifteen or twenty minutes of watching
• The record has to keep pace with the watching
• The codebook you build in two sessions will be assembled almost entirely from what these notes contain

Observation tests your question

• You arrived with a prospectus committing you to directed versus broadcast chat
• Observation will either confirm that split or complicate it
• You may notice a great deal of chat is aimed at neither: replies, arguments, inside jokes between regulars
• The two-way distinction turns out to have a third term
• That is not a failure of the prospectus. It is the cheapest possible moment to learn it.

Before next time

• Topic Selection and Research Questions is due this week.
• The Project Prospectus follows next week. Submit the version you can defend, not the version you love.
• Start your field-notes document tonight: template on the Hub, dated entries, four note types
• Log at least two observation sessions before we meet again, one gaming and one non-gaming
• Read the rest of Chapter 7: manifest and latent content, sampling your observation, edge cases
• Next session we use your notes to separate what any coder can see from what takes judgment