Bulletin report (2020)

Published on by Arnau Siches

Table of contents

This report summarises the publication of the bulletin in 2020.

An attempt to draw some insights from the publication of the weekly bulletins in 2020.

Let's start with some general numbers to frame the rest of the report:

titletotal
Amount of resources read1025
Amount of resources published204
Amount of sources32

Although each bulletin issue has 6 resources, the amount I read per week is higher. A summary:

metricweekly_amount
count36.0
mean28.5
std7.6
min14.0
25%23.0
50%28.0
75%32.5
max45.0

# Resources

I believe all web resources should be accessible via HTTPS so whenever I find a URL using HTTP, I manualy change it to HTTPS. It turns out that some places are either not publishing in HTTPS or doing so but neglecting it (e.g. expired or untrusted certificates). This is the split of resources by either HTTP or HTTPS:

title total
Amount of resources published using HTTPS 198
Amount of resources read using HTTPS 1002
Amount of resources published using HTTP 6
Amount of resources read using HTTP 23

I would've hoped that by 2020 we would've got rid of HTTP but there are a few leftovers (or resisting bastions?).

In the bulletin I also flag whether a resource is a video, a PDF or text as a means to heads up people before the follow a link:

content_type total
pdf 3
text 293
video 10

# Sources

My regular sources are where I find most of my readings. These are the top 10 sources and the total resources I found thanks to them:

source total
lobsters 182
weekrust 70
dataelixir 50
tldr 50
twitter 32
webtoolsweekly 31
dbweekly 23
makermind 22
techproductivity 17
softwareleadweekly 15

In contrast with the top 10 sources for resources I chose for the bulletin:

source total
lobsters 46
dataelixir 13
tldr 9
twitter 8
dbweekly 7
weekrust 6
webtoolsweekly 4
cyberweekly 3
techproductivity 3
makermind 2

So, by the looks of it, the source where I find a resource does not affect whether I will choose it for the bulletin.

Another way of thinking about sources is to look at how often domains repeat. It's a weird one though given that a single domain don't necessarily equate to a single publisher. Good examples of this are github.com, youtube.com or medium.com. In any case, see below the list of domains I read more than 4 times:

domain total
github.com 118
medium.com 45
dev.to 29
nesslabs.com 26
www.youtube.com 13
towardsdatascience.com 8
hacks.mozilla.org 7
leaddev.com 6
jvns.ca 6
untools.co 5
martinfowler.com 5
blog.logrocket.com 5

In contrast, these are the top 10 domains from resources published in the bulletin:

domain total
github.com 27
www.youtube.com 6
nesslabs.com 3
medium.com 3
hacks.mozilla.org 3
www.nngroup.com 2
www.hillelwayne.com 2
untools.co 2
r2c.dev 2
pair.withgoogle.com 2

# Topics

Once I read a resource I tag it with a set of tags that I think represent the topic of the resource. Although I use free-form tags, I try to be consistent.

The top 10 tags in 2020 are:

tag total
rust 185
tool 88
management 58
data 51
database 48
methodology 45
software_development 43
webassembly 43
javascript 41
python 38

And the top 10 that where published in the bulletin:

tag total
rust 33
tool 28
python 16
computer_science 13
data 12
database 11
security 11
sqlite 10
git 9
management 8

Looks like even though I read a lot about management, I rarely find it good enough to be in the bulletin.

# Conclusions

This exercise has shown me a few things I've done right and some others I should improve in 2021. The things to be improved are mostly about recording more information, for example I would've liked to answer the following:

  • How often do I find a resource and leave it for a while to read?
  • When so, how long does it take for me from finding it to reading it?
  • Is there any impact on my reading habits when I'm on leave?

The things I've done well enough:

  • I recorded all resources I've read.
  • I keep all this data as a relational database (actually CSV loaded into SQLite).
  • I keep the bulletin data as a relational database (also CSV loaded into SQLite), so I can merge both databases.
  • I built a rudimentary CLI to manage my resources and another one to manage the bulletins. This makes a huge difference in keeping the management of all this low friction.

Let's see how things change in 2021.