Put phone away box

Week 4 Summary

A bit disappointing (16:45 total), but:

  • This week I had 3 days with extended phone time due to external circumstances (calls etc) and a lot of this growth has nothing to do with doomscrolling. Doomscrolling time remains the same as the last week’s. I reinstalled X to be able to see clearly how much time I spend there. I don’t abuse other doomscrolling drugs.
    • Aside from these days, I had no problem with days of 70-100 minutes total on the phone. This is good. Simply there’s a lot of noise or unpredictability.
  • I have a lot of time logged to Beeminder (more aggressive goal), that doesn’t help to push the limits. Beeminder data is clearing the slot I already fixed - evenings and night time. I can’t use it for other reasons (mornings, work). Let’s keep the commitment as it is.

If I subtract the time I used the phone to communicate or work, I’d be at ~14h, so we can see flat trend around 15±2h a week.

I’ll aim at 12h for next 2 weeks - 100 minutes a day, then go to 10h, then to 1h a day.

If you ask smart LLMs about screen time health, they tell you that 2h a day is healthy for the adult; ask them about the elite performance and they’ll drop to 5-10h a week.

I think it’s worth to aim at 7-10h, to be able to tell the difference.

1 Like

Also, in the past I optimised a lot of life for the phone.

Beeminder goals were optimised for the phone usage to minimise the risk of derail; phone is ubiquitous. I think it could serve as an input device, and not as a protocol. For example, it would be nice to use voice control more and more.

Let me give you some examples - Anki cards should be replaced with voice cards (idk what app yet), Duolingo - just remove it. It can track time I listen to specific podcast or synchronise some statuses from apps I use - imagine you clear your “to read“ backlog on the laptop or tablet, but the phone synchronises the read list via iOS Shortcut.

I am very hopeful for new wearable generations. I don’t see the growing adoption of smart glasses, mass-market smartwatches can’t even recognise my physical writing from the wrist movement etc. No holograms in 2025 still. I think there’s some gap that we fail to fill, that would bridge physical and internet worlds without holding a small portal in my hands.

3 Likes

This is an interesting direction. As I’ve kind of hinted at before, I guess i am tending to go quite a different direction—just using the computer for things instead of the phone (like Anki), since I think it really is meaningfully different. I think I’m kind of prejudiced against all the voice / wearable / etc stuff but I can kind of see how you could use such things without any risk of getting sucked into the phone.

2 Likes

Your approach is correct, because you start from correct assumptions - you reduce the bad screen time, simultaneously leveraging all positive things technology can give to you. I also use laptop more. Phone encourages fragmentation of life. I would like to increase focus time and fight to treat my hobbies or personal projects as first class citizens. Switch from phone or laptop to human interaction would be beneficial long term.

btw: Thanks for participating in the thread, drop comments as often as you wish, I enjoy other perspectives.

I want to say that my plan goes fine, but one day or two can ruin it. I do 1h 30 apart from days where the time spills, I travel or I work with the phone.

I started to use Jomo to block Twitter and I logged out of it in Chrome; there’s only 30 mins of scrolling every day. Next week will reduce to 20. I recommend Jomo.

I have one other comment. My top 5 apps, on a typical day take total of 45 minutes, so I wonder where does the other half come from. If I sum all the apps listed in the settings, it’s like 1h. So where are the remaining 30 minutes? I think there’s some explanation online, or it’s time when the screen is lit? Interesting.

Did someone investigate what the screen time in iOS really is?

1 Like

Week 5 Summary

At 14h we record the lowest total time ever spent, but also a target missed by 3h. This is not a big issue, as the picture is skewed by one day with 3h+.

I still have to figure out mornings. I think there should be simply something that will divert my attention, or some cue to stay aware when the willpower is depleted? Or reframe the idea of the willpower. Looking at the daily charts, there’s this single 20-30 minutes block that I can cut.

This week I had my first day with less than 1h with the phone. I think it’s perfectly doable to go below 10h, I just need more time.

The same chart with a different perspective. Imagine the amount of reclaimed time. I average 1 book read weekly since I started this goal (5 books / 5 weeks, 1 book did not finish).

Also, I like the shift in the structure. Now when I grab my phone and I have my doomscrolling limit exhausted, I’d text some friends on WhatsApp or Messenger. Moreover, this week I spent 1 hour in the Settings app, analysing the patterns and scores.

1 Like

Just wanted to pop by and say I’ve found your thread here motivating. I’m already disciplined about staying off my phone, but I was motivated by the idea that I’d be a monk if I was hitting < 10h on my phone. I like monks.

I don’t mind manual entry goals so my setup is totally boring, but so far I’m very monkish! I might tighten it up since there’s still a few spots I use my phone that I’d rather not (e.g, waiting for kids to get ready for bed), but I can get away with it under these constraints.

Welcome! You are doing great. How do you do it?

I recommend all of you to track as many datapoints possible, take screenshots from the Settings app etc, because now with GPT, we’ve got great analyst to get us some insights. I put the data to gemini.

Total vs. Poking +0.995

Extremely Strong Positive Correlation. This is the foundational insight: Poking Count is the strongest leading indicator of Total Screen Time. Controlling the number of times the screen is activated is the key to overall reduction.

This is correct.

Useful Time vs. Time_logged

The strong correlations, particularly the -0.840 between Time_logged and Useful_Time, highlight a limitation of the phone-box intervention: it’s an “all-or-nothing” tool that restricts all phone usage, not just the non-useful kind. The user succeeded in reducing overall screen time by a large margin (the 31% step-change), but at the cost of the minority of time that was considered productive.

This is partially correct, as reduction was mostly attributed to less driving. “Useful time” was often the CarPlay. But: I moved most of my language goals outside of mobile apps.

1 Like

I’ve been trying to combat attentional problems with internet usage since before the smartphone era, and I was actually literally living as a monk for a year of that time. So mostly I’m just very committed/obsessed/angry about fractured attention and have a lot of habits/motivation/systems for fighting back.

Recently I’ve found the Brick extremely helpful to dehabituate me from recent news-cycle obsessions but I’m not using it right now.

1 Like

Thanks for mentioning Brick, it’s an interesting approach, very similar to my “phone box” idea. Its success (at least as they claim on the landing page) makes me think if my current approach shouldn’t be turned into an app too (“beeminding phone charging time“).

Regarding this:

Yes, basically screen time is all the time the screen is lit via notifications, switching between the apps, CarPlay etc, this is what ChatGPT says.

1 Like

Some updates on tech stack.

This week I am using Jomo with daily 20 minutes hard limit on Twitter. I finally got my physical alarm clock. Also, I set daily goal of 120 minutes inside phone box (update from 90). Minimal time is 45 and maximum is 120.

I think this should get me to 10h unless I have a phone-intensive day.

I didn’t mention I set colors automatically to grayscale between 21:00 and 5:00. Also, a bit of spatial awareness - when I open chrome, messenger or some other apps, there’s 30 percent chance of showing me how much time is left for today - I use Jomo for that too. Thanks @gmzamz for sharing Jomo with us! Remember all: Your threads and posts influence people, share with us so that we can improve!

2 Likes

I’ve got that problem too (also just snoozing) and have yet to solve it :sweat_smile:. I bought a really annoying old style alarm clock with the metal bells, but so far it hasn’t made a huge difference. Maybe I just need to add more points of friction than alarm clock + phone away from bed. I’ve noticed that lots of little frictions is significantly better for dissuading behaviors since they won’t clear the bar for “energy required to submit to the friction > energy required to DISABLE the friction” and will, in the long term, actually stick around. That and setting up a todo in my weekly review to go through and re-enable all the little frictions I disabled during the week :melting_face:

Check this out. there’s a couple of other blogs around that go into it. I decided it was both way too much work and also did not want to buy a mac to access it. But it would be really, REALLY cool to do visualizations with since it’s so incredibly granular

This is the sole reason why I haven’t reverted back to a dumb phone. There’s just too many apps that are either necessary to stay employed or are simply too useful to not have (like maps/etc). I’ve considered lobotomizing my iPhone but the initial friction in setting it up combined with the inevitable occurrence of needing something I didn’t allow-list has put me off. I’ve also looked at buying a not-quite dumbphone, but again, the time/energy/opportunity cost investment is just too high for my current level of screentime problems

T_T I wish I could just read books instead, but it’s probably the #1 reason I am on my phone. Replacing them with physical books just means I lay in bed with a book instead of a screen.

Thanks for the info, I did not realize that. I’m going to have to look into that and if it’s possible to exclude. It’s not an issue now since I am blessed with being able to walk to work but will be in the future when I move

I’m glad it’s working out for you!

Oooh those are good ideas. Going to have to think on what other metrics and proxy metrics one could use for “reduce scrolling”. I jumped straight to time and didn’t really plan much out. Same with all the awesome analysis you’ve done. Going to have to look into doing my own, with the way I log datapoints it should be pretty easy to get some very precise usage info. I usually just swipe away all my screentime related notifications but having it only pop up sometimes would reduce clutter a lot. Could probably even set it up to be proportional to how much screentime you’ve logged.

1 Like

As I read your message I have new ideas.

  1. Don’t use up screen time in the morning - I will do up do-more goal (I don’t have premium now) to log at least X minutes left from Jomo at 9:30 each day.
    So, I have my limit set to 80 minutes a day on Jomo. My Beeminder commitment is 60 minutes a day. So it means that till 9:30 I can use at max 20 minutes of my phone.
  2. Don’t open leech apps in the morning - on opening the app, add current date to a file in a new line. Then, at 9:30 read that file and make sure that the last log is not there from past 9 hours or so.

This should work with do more goals, do less would be more intuitive.

Week 6 Summary

Now I’m at 10h 40 after deductions and almost 13h total :tada:

I anticipated this, because other than a few broken or busy days, past 2-3 weeks have been quite stable.

There was 1 day when I did almost 4h, because I was busy with my phone and used it before bed, something I never did in past 4 weeks. Other than that, I averaged ~1h 30 daily, that means we can safely go for 10h till end of the year.

Also, it’s not really the same time in its structure. Last week, I used X only for 2h. Other top apps: Chrome (2h 15), Messenger (1h 50), WhatsApp (1h), ChatGPT (40 mins). This is basically a quality time with friends and researching with AI, getting things done. Significant part of the chart is a grey block, where the screen is simply lit, but nothing happens (notifications, switching apps etc). Actual, real time spent with the phone is already <10h.

This week I spent also 1h trying to reestablish my iPhone shortcuts after they broke (“database is corrupted”) and I was left with no shortcuts and no automations.

I could try to limit Chrome and ChatGPT, but I don’t really want to. I’m searching for stuff, answering real issues that I have. I think GPT time can be capped at 1h maybe?

I don’t understand phone activations chart, but okay. I think all my strategies so far optimised for getting rid of longer sessions spent with the phone, not really the addiction as in “checking every notification, checking every 5 minutes”. I also miss the definition of screen activation. I think it would make sense to count for a few weeks phone unlocks? Is this even possible?

To recap, some of the experiments overlap, but effective strategies so far:

  • physical separation (from 25h to 15h per week - Beeminder + automations)
  • hard limits on the apps (from 15h to 11h; hopefully to 10h)
  • no-use-before (visibly limits time in the morning, but I can’t give you more specific idea)

It will be clearly visible in the chart. When I started to use Jomo, I started to lose these 15-20 minutes per day. Usage of X and Chrome (where I was logged in to read X) dropped a lot. X went from 2h 50, 2h 40 to now 2h. I can set more aggressive settings in Jomo, but let’s do it step by step.

Beeminder goals:

  • put phone away for X minutes
  • don’t use brainrot app before hour
  • time without phone < minutes before hour

I think about “put phone away sessions”. At the end, my idea is to decrease the number of goals, stay with just one. They feel a bit like crutches, I cant’t imagine living like this for more that a few months.

2 Likes

Next mission to tackle is phone pick ups. Chat says safe level is<50. As you can see I have over 500 pick ups a week. We have to halve this number :face_with_peeking_eye:

The only idea I have is to require 5h a day in a box. Beeminder is not very do-less friendly, it should be something positive. Maybe “day under 50 pickups” boolean daily? Not sure still.

Also I try to think beyond winter, when we spend a lot of time at home. It will be hard to stick to this goal during the summer.

2 Likes

I kinda think the pickups is a case where it is better to just have a do less goal.

3 Likes

Let’s try it then, I set daily limit to 70, this is quite conservative. It’s been between 70 and 90 in the past few weeks.

Next Monday, we’ll try reducing to 50. Let’s see how the weekend will go.

This might be the hardest battle ever so stay in the thread. Finally, a worthy opponent!

I like to write down my own assumptions in this thread because then we can see how wrong we are about the future. So: in my opinion it’s reasonable to get to 30 pick ups a day. The lowest in past few weeks was 50, but I see a lot of patterns related to pick ups when nothing really happens. I’ll cut notifications to bare minimum. I underestimated phone usage in the past, I think it’s really possible to go down to 7h, 5h. This is why 30 pick ups seems realistic.

Now that we see that long chunks of screen time helped to go from 3-4 hours a day to 1-2 hours, I wonder what’s going to be the cumulative effect of 10s-2mins checks few times an hour. Obviously I got a bit weasely now. I developed some sense of phone usage; if I see it’s been just 1h today, I allow myself to spend some time on some app I don’t block (ChatGPT, reading sports news) or use up remaining twitter time. When I see it’s been 2h today, next day I max at 1h on purpose, allowing weekly usage to rebalance itself. This is why I feel the initial goal has been achieved; the metric I had ceased to be a good metric.

It’s going to be a bit more difficult with pick-ups, as they can’t be pulled from any API. It’s easy to do mental math about sitting for 15 minutes with your phone, but much harder with pick-ups. I guess I have to stop carrying the phone around the house as I have a smartwatch.

3 Likes

I think it’s definitely possibly to get well under that because i know I’m way under that with my crazy phone goal. Good luck!

1 Like

Past 2 weeks summary

30.11 - 12h total
07.12 - 18h total

That is because I spent 2 days on conference with almost 10h on my phone - travelling, meeting people and taking notes. Since I was travelling light, I didn’t take a book with me, hence spent some time scrolling and reading things on my phone. I took off hard limit on Jomo deliberately.

Collecting detailed data

That makes me think that I should take outliers into account and understand that there are different types of days in the life. Type 1 - where the reality is under full control, Type B - when something unexpected happen or is planned, and Type 3, or type C - as in Chaos - where all day falls apart due to some events.

In other words, I should perhaps note down screen time and pick ups day by day to understand that 20% of days generated 50% of the screen time, which is perfectly fine, but messes up stats the way I collect them now. I’ll do it with iOS automation and Google Apps Script.

3 Likes

My experience with shortcuts is a bit frustrating. I started to use iPhone as a trigger, but I move as much logic outside of the phone.

Beemind charging time

  1. Go to https://spreadsheets.new
  2. Put in A1, A2, A3: timestamp, type, delta
  3. Extensions → Apps Script
  4. Put this code with proper values
  5. Deploy (make sure anyone can access, authorise new app)
  6. Copy URL. Add ?token=AUTH_TOKEN&action=start or stop
  7. This URL should fire when you use your iOS Shortcut for start/stop charging the battery
const CONFIG = {
  AUTH_TOKEN: 'REPLACE_WITH_SECRET_TOKEN',
  SPREADSHEET_ID: 'REPLACE_WITH_SPREADSHEET_ID',
  SHEET_NAME: 'charging',
  BEEMINDER_USER: 'REPLACE_WITH_USERNAME',
  BEEMINDER_GOAL: 'REPLACE_WITH_GOAL_SLUG',
  BEEMINDER_TOKEN: 'REPLACE_WITH_BEEMINDER_TOKEN',
  MIN_SESSION_MIN: 30,
  MAX_SESSION_MIN: 120
};

function doGet(e) {
  try {
    const token = e.parameter.token;
    const action = e.parameter.action; // start | stop

    if (token !== CONFIG.AUTH_TOKEN) {
      return error('Invalid token');
    }

    if (!['start', 'stop'].includes(action)) {
      return error('Invalid action');
    }

    const sheet = SpreadsheetApp
      .openById(CONFIG.SPREADSHEET_ID)
      .getSheetByName(CONFIG.SHEET_NAME);

    if (!sheet) {
      return error('Sheet not found');
    }

    const now = new Date();

    if (action === 'start') {
      writeRow(sheet, now, 'start', '');
      return ok('Start logged');
    }

    if (action === 'stop') {
      return handleStop(sheet, now);
    }

  } catch (err) {
    return error(err.message);
  }
}

function handleStop(sheet, stopTime) {
  const lastStart = findLastStart(sheet);
  if (!lastStart) {
    return error('No matching start event');
  }

  const startTime = new Date(lastStart.timestamp);
  const diffMs = stopTime - startTime;
  const diffMin = Math.floor(diffMs / 60000);

  // Save raw stop
  writeRow(sheet, stopTime, 'stop', diffMin);

  if (diffMin < CONFIG.MIN_SESSION_MIN) {
    return ok('Session too short – ignored');
  }

  const capped = Math.min(diffMin, CONFIG.MAX_SESSION_MIN);

  sendToBeeminder(capped);

  return ok(`Stop logged, ${capped} min sent to Beeminder`);
}

function findLastStart(sheet) {
  const data = sheet.getDataRange().getValues();

  for (let i = data.length - 1; i >= 0; i--) {
    if (data[i][1] === 'start') {
      return {
        timestamp: data[i][0]
      };
    }
  }

  return null;
}

function writeRow(sheet, timestamp, type, delta) {
  sheet.appendRow([
    timestamp,
    type,
    delta
  ]);
}

function sendToBeeminder(minutes) {
  const url = `https://www.beeminder.com/api/v1/users/${CONFIG.BEEMINDER_USER}/goals/${CONFIG.BEEMINDER_GOAL}/datapoints.json`;

  const payload = {
    value: minutes,
    comment: 'Charging session',
    auth_token: CONFIG.BEEMINDER_TOKEN
  };

  const options = {
    method: 'post',
    payload: payload,
    muteHttpExceptions: true
  };

  UrlFetchApp.fetch(url, options);
}

function ok(msg) {
  return ContentService
    .createTextOutput(JSON.stringify({ status: 'ok', message: msg }))
    .setMimeType(ContentService.MimeType.JSON);
}

function error(msg) {
  return ContentService
    .createTextOutput(JSON.stringify({ status: 'error', message: msg }))
    .setMimeType(ContentService.MimeType.JSON);
}

Log iOS screen time

Similar for Jomo:

  1. Automation at 23:45 daily
  2. Get screentime from Jomo
  3. Ask GPT to generate .gs file you can use to store the data

Tracking update

I lost a few days in a box, because the automation was not logging it, because my iCloud is full and it can’t reliably store things on iCloud when you use mobile data. I don’t want to dive into details, but I think decoupling a trigger and a function or a controller and a model is always a good idea.

On Jomo screen time. It’s different to what iOS reports by sometimes 20%, so I sort of lost continuity of screen time data. This is okay for me, as pick ups become the new north star.

I want to have the awareness of the screen time, but measure pick ups. I’ll leave my phone-in-the-box goal though. I removed all other goals, I find little value in them overall, it’s too much.

More than beeminder goals, I’d like to have a log of my life - did I drink coffee within 1 hour of waking up, did I drink enough water etc. I think this will totally happen before 2030s via some wearable, neuralink or something like that. Oura is close to achieve this dream.

1 Like