Measuring Me Take 2: Doomscrolling as a researcher

Title: Measuring Me (Take 2): Doomscrolling, Measured Like a Tiny System

I’m a CS coterm, and my days swing between deep focus and a lot of mental friction: starting tasks, switching contexts, staring at bugs that won’t budge. When my brain feels full, my thumb knows where to go.

For this assignment, I tracked doomscrolling for a few days and treated it like a system. I wanted to see when it showed up, what seemed to trigger it, and what kept it going. Then I drew two models of the habit ecosystem: a Connection Circle and a feedback loop model.

This writeup includes my log, an ASCII histogram, my two models, what I learned, and what I’d change next time.

What I tracked

Behavior: doomscrolling on my phone.

Definition I used:

• scrolling an infinite feed (Reddit, X, Instagram, YouTube Shorts, news feeds)
• for 5+ minutes
• without a specific purpose (not replying to a message, not looking up something)

How long I tracked:

• 3 days

How often I logged:

• every time an episode ended (event-based logging)

What I logged each time:

• start time
• minutes
• app
• trigger tag (stress, boredom, procrastination, fatigue, notification, autopilot)
• location (desk, couch, bed, transit)
• whether I was avoiding a task

 

My log (3 days)

Day 1 (Mon)

1. 08:52, 7 min, Reddit, trigger=boredom, location=bed, avoiding=no
2. 10:41, 6 min, X, trigger=procrastination, location=desk, avoiding=yes (start coding)
3. 12:18, 12 min, Instagram, trigger=autopilot, location=transit, avoiding=no
4. 14:03, 9 min, Reddit, trigger=stress, location=desk, avoiding=yes (debugging)
5. 16:27, 5 min, News feed, trigger=notification, location=desk, avoiding=yes (writeup)
6. 22:46, 28 min, YouTube Shorts, trigger=fatigue, location=bed, avoiding=no
   Total: 6 episodes, 67 minutes

Day 2 (Tue)

1. 09:11, 6 min, Reddit, trigger=boredom, location=bed, avoiding=no
2. 11:09, 14 min, X, trigger=stress, location=desk, avoiding=yes (hard reading)
3. 13:02, 8 min, Instagram, trigger=autopilot, location=couch, avoiding=no
4. 15:44, 11 min, Reddit, trigger=procrastination, location=desk, avoiding=yes (start project)
5. 18:26, 6 min, News feed, trigger=notification, location=transit, avoiding=no
6. 23:17, 35 min, YouTube Shorts, trigger=fatigue, location=bed, avoiding=no
   Total: 6 episodes, 80 minutes

Day 3 (Wed)

1. 08:36, 5 min, Reddit, trigger=boredom, location=bed, avoiding=no
2. 10:17, 7 min, X, trigger=procrastination, location=desk, avoiding=yes (start coding)
3. 12:55, 10 min, Instagram, trigger=autopilot, location=transit, avoiding=no
4. 14:22, 6 min, Reddit, trigger=stress, location=desk, avoiding=yes (meeting follow-up)
5. 17:08, 5 min, News feed, trigger=notification, location=desk, avoiding=yes (email + tasks)
6. 22:58, 31 min, YouTube Shorts, trigger=fatigue, location=bed, avoiding=no
   Total: 6 episodes, 64 minutes

3-day totals:

• 18 episodes (6/day)
• 211 minutes (~70/day)

ASCII histograms

Legend:

• “#” = 5 minutes
• “.” = 1 minute

Daily total minutes
Mon 67 | #############..
Tue 80 | ################
Wed 64 | ############….

Episodes per day
Mon 6 | ######
Tue 6 | ######
Wed 6 | ######

Minutes by time window (3 days combined)
Morning (06-10) 18 | ###…
Midday (10-14) 58 | ###########…
Afternoon (14-18) 41 | ########.
Evening (18-22) 11 | ##.
Late night (22-01) 83 | #################…

Minutes by trigger (3 days combined)
fatigue 94 | ##################….
procrastination 31 | ######.
autopilot 30 | ######.
stress 27 | #####..
boredom 23 | ####..
notification 6 | #.

Two patterns jumped out fast:

• late night episodes were long
• daytime episodes showed up around starting hard tasks

What it felt like to measure

The act of logging made the habit more visible. I started noticing the first second of it: the tiny moment when my hand goes to the phone before I’ve really decided anything.

I also noticed how often it lined up with a feeling: fatigue at night, friction during the day. Logging didn’t fix the behavior, and it made the behavior feel more predictable. That predictability was useful. It gave me something I could model.

Model 1: Connection Circle (habit ecosystem)

Below is my Connection Circle. The idea is simple: list the parts of the system and show how they connect. Note: no clue why it’s so blurry, sorry.

A few connections I kept coming back to:

• Doomscrolling connects to time lost.
• Time lost connects to stress and guilt.
• Stress and guilt connect to fatigue.

The most important connection in my circle was sleep. When sleep slipped, the next day felt heavier, and the habit showed up more easily.

 

Model 2: Feedback loops (simple causal loop diagram)

For the second model, I used three reinforcing loops and one balancing loop.

Key:
(+) increases
(-) decreases
R = reinforcing loop (tends to grow)
B = balancing loop (tends to reduce)

R1: Reward loop
Doomscrolling (+) -> Novelty/relief (+) -> Keep going (+) -> Doomscrolling

R2: Avoidance loop
Task discomfort (+) -> Doomscrolling (+) -> Task progress (-) -> Deadline pressure (+)
-> Stress (+) -> Task discomfort

R3: Sleep loop
Late night doomscrolling (+) -> Sleep delay (+) -> Fatigue (+) -> Self-control (-)
-> Doomscrolling (+)

B1: Guardrail loop
Friction/guardrails (+) -> Doomscrolling (-) -> Time lost (-) -> Stress (-)
-> Need for relief (-) -> Doomscrolling (-)

These loops helped me see why doomscrolling felt “sticky.” The loops keep feeding the same conditions that make the habit likely.

Main learnings

1. Bedtime mattered a lot.
   The longest episodes happened late at night, and they carried into the next day through fatigue.
2. Starting was a vulnerable moment.
   Many daytime episodes happened right before beginning a hard task: coding, reading, writing.
3. Small cues added up.
   Having the phone within reach made a difference. Being in bed with the phone made a bigger difference.
4. Logging showed me the earliest point of choice.
   The first few seconds mattered more than the last few minutes.

What I’d do differently next time

1. Add a “transition” tag.
   I’d label moments like “after meeting,” “after class,” “between tasks,” “before bed.”
   Transitions showed up a lot, and I want cleaner data around them.
2. Record the stop reason.
   I’d log why I stopped: interrupted, got bored, felt urgency, felt guilt, battery low.
   That would show what actually breaks the loop in real life.
3. Separate quick checks from deep scrolls.
   I’d add one more category:

• “check” (under 5 minutes)
• “doomscroll” (5+ minutes)
  This would give me a clearer picture of how episodes start.

What I’d try next (simple system changes)

I’d pick two changes tied to the loops I drew.

For the sleep loop:

• charge my phone away from the bed
• set a short “shutdown” step: plug phone in, set alarm, then read something offline for 10 minutes

For the avoidance loop:

• use a 2-minute start rule
  When I feel the urge to scroll before a hard task, I do 2 minutes of the next step:
  open the file, write a stub, run a test, outline a paragraph.
  Two minutes is small enough that I usually can do it.

For the reward loop:

• remove infinite-scroll apps from the home screen
• log out of one app so opening it costs extra steps

I like these because they change the environment and the starting conditions, which is where my data says the habit begins.