The Hawthorne effect in your pocket
In November 1924, engineers funded by the National Research Council began a study at the Western Electric plant in Hawthorne, Illinois. Their question was practical. Would brighter lighting on the assembly floor make workers produce more relay switches per hour? They raised the lights, and output went up. They lowered the lights, and output stayed up. They lowered them further, until the workers were assembling switching equipment by something close to moonlight. Output kept rising.
The illumination studies ran for two and a half years. Four further sets of experiments followed at the same plant until 1932, the most famous of which observed five women in a Relay Assembly Test Room while researchers altered their break schedules, hours, and pay. The story that came out of this body of work, and that was retold to generations of management students, was simple. Workers produced more whenever someone was watching them, regardless of what was being changed. The sociologist Henry Landsberger named the phenomenon the Hawthorne effect in his 1958 book Hawthorne Revisited, and it became the textbook example of how observation alters behavior.
The textbook version is partly wrong. The phenomenon underneath it is not.
The reanalysis
In 2011, the economists Steven Levitt and John List published a paper in American Economic Journal: Applied Economics asking, in their title, whether there had really been a Hawthorne effect at the Hawthorne plant. They had recovered original data from the illumination experiments and reanalyzed it with statistical methods the original researchers had not had access to. Their conclusion was deflating. The effect, properly measured, was small and confined to particular conditions. The popular version, in which productivity rose any time workers knew they were watched, did not survive the data.
Earlier critics had reached similar conclusions through different routes. John Adair’s 1984 review in the Journal of Applied Psychology had already argued that the Hawthorne effect was largely a methodological artifact, more useful as a cautionary tale than as a measured phenomenon. The textbook version, it turned out, had been a product of mid-century retelling rather than careful reading of the original numbers.
What survived the corrections was the broader fact, not the specific story. Being observed does alter how people behave. The phenomenon has a more precise name in the methodological literature: reactivity. It shows up reliably in studies that ask people to record themselves.
Reactivity in self-tracking
Asking participants to log their meals reliably changes how they eat. Asking them to count their cigarettes reliably reduces the count. Asking students to record study hours produces more study hours. The 1999 review by Korotitsch and Nelson-Gray in Psychological Assessment surveyed dozens of these studies and described self-monitoring reactivity as a robust enough effect that clinical psychologists use it as a standalone behavior change technique. You do not need a treatment plan. The act of recording is the treatment.
A 2016 meta-analysis by Harkin, Webb, Prestwich, and colleagues in Psychological Bulletin pulled together 138 studies covering nearly twenty thousand participants. Prompting people to monitor their progress toward a goal produced a medium-sized effect on the behavior they were monitoring, with smaller but reliable effects on goal attainment itself. The effect held across health behaviors, study behaviors, and weight management. Self-tracking does not just describe behavior. It moves it.
For someone trying to lose weight or quit smoking, this is exactly the property they want from a tool. For someone trying to keep a record of what their life is actually like, it is the property they want to avoid.
The diary as a measurement instrument
Most diaries are self-monitoring instruments, whether or not their owners think of them that way.
A gratitude journal asks the writer to list three things they were grateful for today. The day was always going to be lived. From the moment the journal exists, the day is also being scanned for moments worth being grateful for, because the writer knows the question is coming at bedtime. The day-as-lived and the day-as-scanned-for-content are no longer the same day.
A mood tracker asks for a daily score. Studies of ecological momentary assessment, the research practice of pinging participants for in-the-moment mood ratings, have repeatedly found that being prompted to rate one’s mood changes the rating, particularly when prompts are frequent. The numbers describe a person who knows their mood is being asked about.
A bullet journal, a five-minute journal, a free-writing practice: each of these is an active intervention in the day, not a transparent record of it. The person keeping the journal is, in part, the person the journal is shaping.
This is not a critique of those practices. They are designed to change the practitioner, and the evidence is that they do. It is a point about evidence in the other direction. The diary kept through any of these methods is a record of someone who keeps a diary using that method. It is not a record of who the person would have been if no one, including themselves, had been watching.
What does not get observed
Reactivity requires awareness. The workers at Western Electric knew they were in a study. The smoker counting cigarettes knows the count is for the count’s sake. The journal keeper knows the page is waiting at the end of the day.
The category of behavior that escapes reactivity is behavior produced for some other reason, observed afterward through traces it left behind for that other reason. A calendar event was created because there was a meeting. A commit was pushed because the code was ready. A check-in was logged because the user was at the place. None of those decisions involved a diary as audience. Each was complete on its own terms before any diary existed.
This is the ordinary state of digital life. Most people produce more such traces in a day than they realize. The calendar holds where they went and who they met. The task manager holds what they finished. The repository holds what they built. The chat client holds who they spoke to and roughly about what. The music service holds what they played at what hour. None of these traces was produced for a diary, which is why none of them is bent by a diary’s gaze.
What deariary collects, and what it does not ask for
deariary assembles a daily entry from these traces. It connects to thirteen tools, including Google Calendar, GitHub, Slack, Steam, Todoist, Bluesky, Toggl Track, Discord, Swarm, Last.fm, Trakt, and Linear, and pulls activity each tool already records. The product does not ask the user to fill in a mood, score a day, write a reflection, or choose three things to be grateful for. The user is not the source of the data, in the sense that no part of the user’s day depends on the diary’s existence.
The reactivity surface is small, and the smallness is structural. There is a moment when a new user starts naming calendar events more carefully because they know the events will appear in their diary. This is real. It is also the same drift that happens when anyone starts using a calendar at all: events become more legible because legibility helps the future self, regardless of who is reading. The diary did not create that drift. It joined it.
What is absent, by design, is the daily ritual of recording oneself. There is no checkbox at 9 PM. There is no question that has to be answered about feelings. There is no daily metric that can be gamed by going for a quick walk before bed to bump a number. The day arrives, the data flows, the entry appears the next morning. The day described in the entry is the day that would have been lived anyway.
The hardest record to make
The deepest version of the Hawthorne problem is not that observation changes behavior. It is that any system designed to record a life will, by recording, become part of the life it is recording. This is a property of attention, not a flaw of any particular tool. Whatever a person puts in front of themselves daily becomes a thing they adjust toward.
The only partial escape is to use traces produced for reasons that have nothing to do with recording. Those traces existed before any diary saw them. The diary that uses them is, in some sense, not watching. It is reading what was already written by the day, in the language the day uses with itself.
This is not a perfect record. It is not even a complete record, because plenty of life happens away from any tool. But it is the closest available to a record made of the person who was actually there, rather than the person who was deciding what to write down.
