I've Been Tracking Everything in My Life for A Year

In short: For a year, I tracked every action in my life alongside my well-being. I then applied Bayes' Theorem to understand how my choices and actions shape my well-being.

But Why?

Firstly, I realised that I could lead a less passive life, one where I’m not on autopilot, being tossed around by turbulence and near misses. Ultimately, it’s my responsibility to actively engage each day, steadying myself to live according to my wants and expectations.

Secondly, the concept of a learning journal intrigued me. The idea that each previous day could lend its knowledge to the present was fascinating. I wanted to test if it was possible to translate such data into actionable knowledge and lessons that I could use.

What I did:

I tracked my well-being - mentally, physically, and socially - every day. This allowed me to evaluate the daily actions I tracked against these records, creating a constantly evolving feedback loop for my vision. All of this was contained in what became a data-centric journal that I developed for myself.

From “To Do” to “Done”: How I tracked my life

How Habitually Works

Habitually focuses on what has been accomplished. The user interface illustrates the answers to the question of what has been “done” and by how much on any given day. Only actions that are “present” will be visible, while the rest are tidied away.

The Data Structure

Every action is represented by a nested map. Here’s the hierarchy:

20231004{

    1696408920000: 11.0,

    1696410049833: 3.0

},

20231005: {

    1696488060000: 7.25

}

Despite minor redundancy, this structure allows for quick searches and easy sorting. It’s especially useful for sorting a list of actions into “present” and “absent” for the UI.


Should Location Data Be Included in the App’s Tracking? 

Location data could potentially provide insights into daily routines and habits. However, it’s important to consider whether location should be included in the tracking. Could it not be viewed as a dependent context that surrounds actions? Moreover, its inclusion could complicate self-tracking and reflection. While location data can be informative, it’s important to ensure that it doesn’t cloud the primary purpose of the app, which is to track and reflect on my actions.

How did I track my well-being?

I used three questions involving smiley faces and a 5-point Likert scale (Bad, Displeased, Neutral, OK, Good) to track my well-being. My aim was to keep the experience as simple as possible and minimise over-analysis or discomfort with in-depth self-reflection. While individual data points may not be 100% accurate, they offer insights over time, with minimal focus on any single day.

I experimented with how I could frame the questions I asked myself. For now, I have tried to create one present reflection and one future reflection, but I will probably revisit this at a later date.


Reflecting on my Well-being Using Bayesian Statistics

During the early days of development, while going through the backlog of Veritasium videos, I watched “The Bayesian Trap”. The video discusses how our own internal Bayesian abilities are flawed and concludes that we should check the priors (opinions) of our own daily lives. As one commenter put it, “Came for the mathematical insight, stayed for the existential crisis.

Determined to use this concept in my app, I saw its potential for evaluating my life and, in turn, taking action to get my house in order. Better yet, all this unbiased ruminating can be left to the app. With well-being data being provided daily, it would gather a more accurate picture of my life in the same timeframe. Relaying back the likely impact my actions have on me, giving me the chance daily to reflect and respond to these now updated priors.


My current implementation

In my current implementation, I use a Naïve Bayes Classifier. I define an “instance” as any given day, the “categories” as the well-being outcome for that day, and the “features” as every action I am tracking.

A unique challenge arises due to the evolving nature of the data: actions can be added or removed at any point, and additional instances are created daily. This requires the features to be flexible. For each feature, I incorporate three states:

In its current state, only present features are being learned and classified. Considering it is a “Done” journal, this is the evaluation in its most basic form. However, adding absence and existence into these calculations remains an open question but requires handling multiple cases surrounding infinitely changing data. Mainly, what needs addressing are:


My Struggles with Actioning Well-being Insights

With an autopilot of my own, everything should have been smooth sailing. However, the turbulence has never felt more painful. Despite being more informed than ever, I questioned why I wasn’t smart enough to avoid flying into storms. This led me to a nurture vs nature debate, where I wrestled with the question: Can you change your identity? I wondered if I was simply priming myself to go against my nature and, in doing so, causing more harm.

There isn’t much discourse when it comes to experimenting with change. Listicles or blogs often suggest that you stop thinking about your wants and simply choose an action and work on it. Alternatively, they propose that instead of focusing on goals and aspirations, you should choose the processes of a successful life and live it.


Delving into the Grey Area 

What often falls short is the belief in binary good and bad, something that my application had inadvertently reinforced. Little thought is given to figuring out why, how much of something, and in what context, to achieve the desired balance in our lives.

As someone who struggles with inaction and being frozen, even when I intellectually understand “what is good and bad,” this gap between knowledge and action felt like a major roadblock. I believe it devalues, at least for me, the categorisation of good and bad. This is just the tip of the iceberg of mindless living and controlling actions. I can remove every tip, but the foundations that make up a poor action will always persist.

My initial approach to solving the ‘why’ and ‘how’ of each action has been through:

So, What Did I Record?

My data-driven journal, as it has evolved over a year’s use, contains 145 tracked actions at the time of writing this article, and that list is sure to grow.

In a Gantt chart style, it visualises not just every action in my life, but also the date I began tracking it, and when I performed it.

There is an undeniable thrill in knowing so much about myself. Even though I am aware of missing pieces, a lot more than I could ever have predicted has been brought into the spotlight. With a sense of continuing this indefinitely, I have most definitely caught the quantified-self bug.

My actions can be categorised as follows:

We all possess a certain level of abstract understanding about who we are. However, it’s rare to delve into such detail. For me, the need to distinguish YouTube from general Media Consumption was eye opening.

I see the number of verbs being added gradually decreasing to zero. Until that happens, I am unable to predict a definitive or near-final tally of actions.

How Have I Been?

Given that 75% of my activities are tied to media and entertainment, no wonder that my mental well-being leans towards the positive. I could conclude now that I’m sacrificing my social and physical well-being for short-term mental satisfaction.

All Time Well-being

Apr - 2023

May - 2023

Jun - 2023

Jul - 2023

Aug - 2023

Sep - 2023

Oct - 2023

Nov - 2023

Dec - 2023

Jan - 2024

February - 2024

March- 2024

Putting the same data onto a line chart makes revisiting those moments of progress until the almost inevitable crash a less enjoyable experience. It’s a gut punch to witness my struggles associated with managing my physical well-being (something I value a lot).

Stats and Findings

My Actions as a Whole

Since starting this article, I’ve added 13 more actions to my journal, bringing the total to 158. The clear bias of my actions towards mental prosperity, at the expense of average social well-being and negative physical well-being, highlights a drastic need to rebalance my priorities across different dimensions of my life.

This is perhaps the most fitting chart to illustrate my need to look beyond happiness. There are clear issues lurking beneath the surface that would temporarily reveal themselves from time to time.

Developing Habitually

Its only appropriate that we start our journey of oversharing with my work on developing Habitually. From the initial MVP, all my life has since been recorded.

Further122Sessions of Development

Developed Every 2.53Days

Longest Streak 13 Days

Average Streak 3 Days

It’s fascinating to observe the natural phases of development, testing, and debugging from the charts, especially the year view.

Any peak represents the floodgates opening for creating features that I’ve jotted down or reshuffles in the app. Meanwhile, quieter months are spent researching how the app could grow and evolve, working on any debugging that would be flagged up by crashes or misbehaviours.

Insights from the charts could be taken away on either a micro or macro level. On a micro level, developing Habitually is simply a negative physical influence, and it’s perhaps truly one of the worst things in my life. However, a deeper and more thoughtful explanation is how I am inherently more likely to enact better mental health options but make more unhealthy physical well-being choices, among other things.

YouTube Observations

2286 YouTube Videos Watched

To be honest, I’m not sure how to comment or react to this. So, let’s reframe it: if I had read one page from the LOTR trilogy for every video I watched, I could have completed it twice over. Maybe that could be an experiment for next year.

I’ve watched nearly 1500 videos across 14 channels. The scary part about the social media algorithm is that all of these were recommendations I would never have sought out on my own. It’s startling to realise just how much YouTube knows about me, enough to guide me to watch these specific channels.

Interestingly, six YouTube channels had a positive impact on my well-being across the board. These include:

Upon further analysis, these channels are among the least watched in my list, averaging at 17 videos across the year.

Traditional Entertainment Observations

If I were to rewind the clock back five years, my consumption of gaming would have been significantly higher. Going through university, most, if not all, of this time was gradually transferred to other forms of entertainment.

LIVERPOOL FC

28

WINS

9

DRAWS

4

 LOSSES

46 Movies 
1018 TV Episodes

I’m relatively slow to pick up on new TV shows, with fewer than six shows that I’ve started watching in the last five years. From one-off viewings, a lot of hours seem to slip through the cracks. Soon, I plan to implement a generic TV tracking action for any show that I’m not 100% committed to. I’ll define these as shows that I’ve never binge-watched in succession or made an effort to re-watch.

I’ve come to realise that it’s a trap to believe that the best and only course of action is to reduce my entertainment consumption. No doubt, it contributes to the neglect of my physical and social well-being. However, I could also alter my habits of consumption. Before going cold turkey and moving to the endless healthy and social activities available, why not transform a weakness into something of greater value? Similar in vain to the LOTR comment.

Sleep

Average Sleep 2023
6Hrs 22Mins
Average Sleep 2024
6Hrs 47Mins

Last year, I noticed on multiple occasions that my sleep was far from the recommended amount. As a result, I set a target of six hours per night. It’s encouraging to see that I’ve achieved this target on nearly 70% of days, and that percentage is rising.

Upon further research, I’ve adjusted my sleep goal to six hours and forty-five minutes. My achievement rate for this new target is closer to 40%.

Note to self (group values in the pie chart)

Its interesting to observe that I’ve lost a total of four days’ worth of sleep from waking up on Fridays and Saturdays. Considering the small difference, it seems to be a travel-related factor where I’m more active and require fewer hours of sleep on these days. With it potentially being recovered Tuesday and Wednesday.

This observation has inspired me to model the quantity of my sleep against my well-being.

Something that this has inspired is modelling the quantity of my sleep against well-being. Surprisingly the trends found are insignificant, the amount of sleep I get essentially has no say on my well-being on any given day. With so much information regarding the importance of sleep I have nothing of value beyond an insight that the amount of sleep in my life is overshadowed by other more impactful factors.

Surprisingly, the trends found are insignificant. The amount of sleep I get essentially has no impact on my well-being on any given day. Despite so much information emphasising the importance of sleep, I’ve found that the amount of sleep in my life is overshadowed by other, potentially more impactful factors.

Another aspect of this relationship is the impact of the time I wake up on my well-being. This can be inferred from the timestamp I record my sleep data on.

Considering the much stronger trendline fitness explaining more of the variation in well-being based on the time I woke up, could it be argued that this correlation is inherently more causal? Waking up earlier correlates with more activity and less sloth-like behavior. Would forcing myself to wake up earlier with no plans further investigate this?

For a little more accuracy and understanding, here are the findings of a regression analysis of my sleep behaviour on my well-being:


(NOTE)

I have left the results for the regression analysis above that can be expanded and collapsed as I am not 100% confident in my understanding of them.

Web & Social Observations

News
Longest Streak 273 Days
Avg Streak 19.4 Days
Avg1.15 Days Between
1 Daily Avg Visits
Hacker News
Longest Streak 13 Days
Avg Streak 4.25 Days
Avg1.29 Days Between
1.76 Daily Avg Visits
Reddit
Longest Streak 66 Days
Avg Streak 8.86 Days
Avg1.14 Days Between
20 Daily Avg Mins

Distraught by the realisation of my daily reliance on the news, I began skipping days of news consumption in November. On some days, I consciously ask myself if I truly need to read the news, while on other days, I unconsciously skip it outright.

Reddit Usage

In addition, I want to focus on my Reddit consumption, which was always recorded post-consumption, unlike the other two. This means that after using Reddit, I would refresh the screen time use of the app and add it into the journal.

Could this usage accountability be the driving force for my unconscious reduction in my usage of Reddit?

Across the board, going beyond 20 minutes has more negative impacts on my well-being. Is my unconscious reduction because I had innately understood the diminishing returns of Reddit and became a lot more utilitarian about my usage?

Food Observations

For the cereals listed, I adhered to the nutritional guides, considering 30g as a single serving. Crunchy Nut, which contains 8g of sugar per serving, for me amounts to 2kg of sugar in a year. Adding Rice Krispies into the mix increases this to 2.5kg of sugar annually, all from a meal that is generally not very filling.

From my personal records, I know that the last time I visited the gym was June/July. Having achieved my fitness goals for the year, I decided that further gym visits were unnecessary based on my daily activity. It’s now clear that without the accountability provided by regular exercise and increased dietary awareness, an unhealthy habit can easily take root.

Rice Kripies
Crunchy Nut

As previously mentioned, I find it challenging to utilise this information effectively. Only through introspection have I arrived at the conclusion that I want to reduce my cereal consumption. More importantly, I aim to be less rigid in my choices, opting instead for experimentation with a wider variety of options.

Selfcare & Work Observations

Haircut
Longest Streak 1 Day
Avg Streak 1 Day
Avg 23.5 Days Between
1 Daily Avg Cut
Longest Break 62 Days
DIY
Longest Streak 9 Days
Avg Streak 2.64 Days
Avg 2.35 Days Between
1 Daily Avg Visit
Longest Break 16 Days
Wash Car
Longest Streak 2 Days
Avg Streak 1.25 Days
Avg 37.56 Days Between
1 Daily Avg Wash
Longest Break 153 Days
Cardio
Longest Streak 14 Day
Avg Streak 6.71 Days
Avg7.21 Days Between
22 Daily Avg Minutes
Longest Break 245 Days
Gym
Longest Streak 1 Day
Avg Streak 1 Day
Avg 13.11 Days Between
1 Daily Avg Visits
Longest Break 238 Days

Reflecting on my cardio and gym stats, I see parallels with other aspects of my life. Just as I maintain regular haircuts and engage consistently in DIY tasks.

These stats, much like my haircut, DIY, and car washing routines, should not be viewed as transactional or burdensome tasks, but rather as part of a maintenance and upkeep routine that contributes to my overall well-being. This is a shift from my previous approach, where I would go to the gym six days a week until I achieved a thinly veiled goal.

Reflections and Improvements

In the lead-up to writing this article, I must confess, I cheated.

Maybe in planning for this section, I started a side journal in a note-taking app. There, I quickly captured any thoughts and feelings related to what I was doing. This was mainly to prevent these final sections from devolving into incoherent ramblings.


Technical Reflections

Before I highlight the current shortcomings of Habitually and suggest solutions, I find it necessary to unambiguously define the two main pillars that Habitually comprises of:

A majority of the work in the previous year went towards the first objective, and I am more than satisfied with the UX. However, there are a few additions or tweaks to be made to which I am open to suggestions and look forward to. As well as some obstacles to overcome, mostly surrounding the second objective. 


Misinterpreting Likelihood as Fact:

While it wont go away, the current arrow-based impact signals are, at most, worth being considered as coloured flags. One of my biggest pet peeves is the determination of things as good and bad. And, it did not take too long for my BS detector to take notice of these signals.

Not only can tweaking calculations influence outcomes, but with infinite data and possibilities, these signals only function to test our own expected priors (outcomes) as they evolve. I had hoped to sway its influence a little more by implementing relative rankings in hopes of increasing fluidity in the categories. However, this was not enough. Hence, the solution I began producing in October was:

Reflecting on emotions and behaviours surrounding each action, mainly through the “Introspection of the Day” feature. Here, every action is given the same question which would prod into the why and how, essentially beginning a conversation surrounding my actions.

Soon, I would also like the “Introspection of the Day” to save my responses to each question as they come, allowing for not just multiple answers to the same question but an ability to observe my answers over time. This would provide greatly needed context and insight into my behaviour and feelings regarding these flags.


Data Quality

Not all tracked data is relevant to learning, and this affects how actions are measured. For example, my tracking of sleep duration is a daily entry, inflating its influence on the model even though it is a non-actionable data point. This is also one of the strongest cases right now against adding location tracking.


For me the initial solution will be to add a opt-in button filtering out from learning any actions that have opted in. Also, if I was to implement some filter prior to learning this could be an avenue for location tracking to be added as I could outright deny any location related context being learnt on.

My initial solution will be to add an opt-in button, filtering out from learning any actions that have been opted in. With this potentially being the way location tracking being implemented.

Lack of Supporting Models and Analyses

Only through this exercise has there been a recognition of a lack of models or charts to showcase deeper relationships within the data. From within my own report, I have identified a need for categorising the data. Without it, I may have never truly become aware of my content addiction, and larger macro trends in my life that only come from relationships beyond a single action and my well-being.


Lifestyle Reflections

Here, I aim to reflect on the role Habitually has played in my life and how it has transformed my lifestyle over the course of a year.

Creating this article has been a journey in itself. The phrase “I don’t know what to write” has often surfaced, sometimes as a placeholder, sometimes as a serious opener. This being my first article, I could describe the entire experience as a test of confidence, and optimism. At times, I’ve struggled with the realisation that I’m just at the beginning of a journey to understand who I am and what I desire.

Never before have I had such extensive knowledge of my personal history. Which is at odds with how I feel my actions and life is boundless and fluid in my head. Its as if since delving into the intricacies of my life has made it harder to analyse why I am who I am and not some multi-dimensional variation. While I’m learning to move beyond regret, reflecting on my past has brought forth my own insecurities on my situation. My actions and identity have closed many doors, some of which may never open again. But I can no longer dwell on the “ifs” and must focus on what’s next. There are still infinite doors open, and even backdoors to those I once considered closed.

It may be maturity, but I now believe that my current situation is probably where I’m meant to be. I’ve started to reel in the possibilities of what I can do and want to do. By understanding my present state and location, I can prioritise what truly matters to me. A 16-hour day full of infinite choices is meaningless; only the weight of actioning them is meaningful.

Since starting this project, I’ve found a way to contextualise all my actions and purpose. I haven’t always appreciated what’s right in front of me, often seeking greener pastures and neglecting my own. By emulating the scientific approach to my life of —observation, hypothesis, experimentation, repeat—I hope to achieve my wants and expectations. I don’t want to delude myself into thinking I have full creative control, but being mindful enough to monitor my complacencies and believe in myself is a significant step forward from where I started.

Most importantly, from my observations that life doesn’t magically transform within a year. Change is often slow, and setbacks and backsliding are inevitable during challenging times. It may seem appropriate to average or distribute my well-being stats to evaluate my progress, but I believe that’s misguided. There’s never a “correct” time to determine whether you’re happier or healthier etc in relative terms.

Ultimately, it’s strange to recognise all this information and even stranger to feel indifferent about my actions. There’s a lot of shame and chastising online about people’s lives, but the only course of action is compassion, not self-hatred. I can’t hate my consumption of YouTube, especially if it’s an action I continue to make. What I can do is question whether I should watch this video now and why not at some other time?

Conclusions

I want to start with a massive thank you for getting this far. This piece has occupied my thoughts for several months. Initially, it was intended to join the annual productivity and life-hacking discussions that take place in December and January. However, I struggled with its purpose and development.

Perhaps in a year’s time, I could analyse what change truly looks like. But for the year just passed, I feel not only accountable but also, for the first time, supported and confident about my life decisions. I would describe this as emotional maturation, as my understanding of my thoughts and feelings has gradually increased.

While I’m still prone to making short-term, impulsive decisions, being aware of them has significantly reduced their frequency compared to before this experience. I can now let go of self-loathing. I have a clearer understanding of what a balanced life looks like for me.

I like to think of this journal as a GPS, similar to Google Maps. It provides me with all the relevant information about the journeys I want to undertake, but it never takes control. I may deviate from the route, but I can always find my way back. I can choose to speed up or slow down, but it will continuously guide me.

Revisiting the iceberg metaphor from earlier, we often focus on the few visible actions in our lives, labelling them as “good” or “bad,” with little regard for the underlying behaviours that extend far deeper. The healthiest way to start uncovering what lies beneath is not to immediately start hacking your lifestyle or focusing all your efforts on a few aspects. Instead, quickly track and note your daily activities. A rough estimate suggests that I spent an average of a few minutes daily on tracking. Even if that doubles to four minutes per day, a year of journaling amounts to just 24 hours.

Eventually, you can start to contemplate the actions you’re taking, with a better understanding allowing for more thoughtful action. Be intentional with this introspection and self-evaluation, like setting aside a day in your calendar for it. It was months before I began to recognise any deeper patterns and relationships, with this article perhaps being my first serious attempt.

To quote Derek from Veritasium, “Our actions play a role in determining outcomes, determining how true things are.” So, in a similar vein to his closing statements, ask yourself: What areas in your life are you curious about?

What has been most valuable is realising that almost no advice can beat a series of important questions, feelings, and thoughts that you can ask yourself. What areas in your life are you curious about? What actions can you observe? And is your opinion or hypothesis on this action backed up by any hard truths?

© Habitually 2024 Privacy Policy Press Kit