Self-research on smartphone use / withdrawal

Hi, last week I shared at the Keating Memorial event a self-intervention I did during May to try understand better my relationship with the smartphone. Basically gathering reactions with a one button tracker and in parallel with annotations for the most relevant events during the “absence” of the smartphone - and my kind of user experience switching to a dumbphone. I also share here my slides in case someone is interested or doing similar things.

Enric digital minimalism intervention - May 2021.pdf (1.5 MB)

One of the questions I think from @Steven_Jonas was about rebounding effects after such type of “withdrawal” intervention, which I had no time to answer properly then. I think this is very important for what I’m trying to do in further approaches to the topic, so what I’m figuring out how to do more systematically now is a combination of similar tools adding Activity Watch (for the periods when I’m back to using the smartphone as usual).

For example initial data from Activity Watch for July (when I was several days on holidays travelling and trekking, with my SIM back in the smartphone as usual) is that it was very intense, coinciding with my own impressions, but I still don’t have data to compare with more regular periods - when working from office / home, in usual places, commuting, etc. However, I wonder if some of the stats of smartphone use (although not including laptop use) during that month could indeed reflect a sort of rebound effect to some extent - I attach them too as a screenshot.

So any thoughts or advice on this welcomed too :slight_smile:


Hi Enric, During the Weekly Self Research Chat we talked about your use of the Puck.js as a One Button Tracker, and the problem of false positives and how to possibly fix this by using a DIY printed case that recesses the button. Is it correct that you think the false positives were caused by pressure on the button in your pocket? Can you say a bit more about this, how did you know there were so many false positives?

Hi Gary, thnx for the question. The reason why I think the puck.js (or the use I did of it) was not reliable is because when merging the .csv files from different weeks, I realised many very short and continuous (just by seconds) presses that I was sure were not intentional and did not match my memory nor my notes. I copy & paste a sample at the end of this message where it’s easy to see what I mean.

I assume these false positives were caused by having the puck in my trousers pocket all the time, and therefore by being very easy to activate by mistake, just by moving around, etc. Maybe some of the additional data (there’s one for temperature I think) could bring more light on this. In any case, I also learnt by practice on the importance of calibration beforehand :slight_smile: Part of starting this was somehow impulsive and I should have prepared, tested and checked better this kind of things first.

Thu May 13 2021 17:49:46 GMT+0200,-,0.29354858398,23.625,0.09721517562,-740,5468,5839
Thu May 13 2021 17:49:47 GMT+0200,-,0.25238037109,23.625,0.11436939239,-114,5443,3555
Thu May 13 2021 17:50:25 GMT+0200,-,NaN,23.625,0.09920962651,-3596,-2610,7023
Thu May 13 2021 17:50:27 GMT+0200,-,1.14544677734,23.5,0.11348962783,5212,3824,4966
Thu May 13 2021 17:50:29 GMT+0200,-,0.4419555664,23.625,0.10934257507,-383,5688,5681
Thu May 13 2021 17:50:29 GMT+0200,-,0.15661621093,23.625,0.08769512176,-168,5739,5572
Thu May 13 2021 17:50:32 GMT+0200,-,0.88952636718,23.625,0.11453104019,-3536,3940,5965
Thu May 13 2021 17:50:33 GMT+0200,-,0.74731445312,23.75,0.1111793518,-3437,3606,6141
Thu May 13 2021 17:50:35 GMT+0200,-,0.64813232421,23.75,0.09898328781,-3318,3531,6319
Thu May 13 2021 17:50:37 GMT+0200,-,0.55047607421,23.75,0.10983467102,-3149,3768,6352
Thu May 13 2021 17:50:39 GMT+0200,-,0.49011230468,23.75,0.11293601989,-3482,3932,5818
Thu May 13 2021 17:53:24 GMT+0200,-,0.80361938476,24,0.43676153818,1634,-5303,5832
Thu May 13 2021 19:04:17 GMT+0200,-,0.5080871582,22.625,0.10237979888,6943,-3157,-4816
Thu May 13 2021 19:25:09 GMT+0200,-,0.83148193359,22.75,0.10321172078,-3473,8145,-177
Thu May 13 2021 19:25:14 GMT+0200,-,0.81726074218,22.75,0.10769716898,-7560,-1848,2881
Thu May 13 2021 19:25:17 GMT+0200,-,1.42849731445,22.75,0.10251410802,-7161,-578,2715

Do you think the key issue is simply the button is not recessed?

Sorry for the off-topic post, but I got intrigued by the Puck.js. Looks like an amazing tool for self quantifiers. I am interested especially in the tracking of events by button presses. How does this work? Is the puck.js able to store events with timestamps in an internal memory?

The reason I am asking is because I had tried something similar in the past using rf buttons to store events in a database. However, a limitation of this approach is, that it only works around the house, where an rf-bridge is available to receive the signals. So the puck.js would offer more flexibility in this regard.

1 Like

I’m more or less obsessed with the one button approach, I’ll start a new topic in the next day or two and link it from here.

1 Like

Yes basically, that is one of the functions, to save in internal memory .
Check the following page: Cheap & open source one-button tracker

You can also record for example when a door is open or closed based on the magnet field when attaching it to a door frame.

1 Like

Hi, great @Agaricus if you open a specific topic for the one button. I could share there results from a basic thing I’m trying this week, to try reproduce what complicated my intervention with the puck.js: just having it in one of my jeans pocket, without pressing it intentionally at all, and let’s see if it records anything accidentally. Because I don’t think the issue is that the button is not recessed after intentionally pressing it.

I guess @datata @coder1989 I could share in more detail there my experience with the tool, instead of this thread, but as a summary: I only used the timestamp and duration data stored on the puck (not the rest of it), which helped me overall with the need I had to track perceived events for my intervention. However, besides these false positives issue (and derived cleanup) I also lost some of the data from the last days when downloading it via bluetooth to my PC using @gedankenstuecke hack at Puck.js One Button Tracker - I guess this is another little but relevant issue to solve :slight_smile:


So you just tracked the positive/negative and then in the cases that you took notes manually try to link it with the handwritten comments right?

I am thinking to use it for multiple habits tracking or emotions journal. but for that I would need different click combinations for different measures.

Maybe a long press and then make use of the colors to select the relevant one.

I’ve created a thread with background and invitation to discuss all things One Button here:

One Button Tracker.

@datata Several people have done projects involving multiple presses to record different phenomena or different intensities. The prototype I use also records duration and intensity of the button presses, but I’ve never used that feature. Enric did use duration, however.

Right @datata for that first intervention my handwritten notes had also the negative / positive / reflection category when I wrote them on spot, in order to later on match them with the pressings of the one button tracker. I think in a very similar way to what you plan with these habits / emotions tracking, the merging of data on a spreadsheet required to assign then these categories in an additional column based on my notes - matching usually the timestamps with the longer duration of pressings (more “significant” than the shorter ones, which were negative by default reflecting dependence, discomfort, fomo, etc).

1 Like

I haven’t implemented it for the Puck.js one-button yet, but someone made a little function for the Puck that can differentiate between short & long presses and allows to combine those presses into sequences (e.g. double click with 2 short presses triggers another action than a single short press). This could be a potential solution as you wouldn’t need to post-process the data as the click combinations are marked right away. I’m currently not having a Puck.js around to play around with it though :slight_smile:

1 Like