hollerith: logging social behaviour

• hangouts provides searchable chat logs (in gmail), therefore the precise number of people contacted on a day can be known.
• turning off conversation view in gmail settings expands chat logs into one message per entry.
  • this could be construed as a vague measure of the volume of chat.
  • it is further possible to see how many of these messages were sent and received (there is a small » indicating who sent the message)
    • I believe that sent:received must be close to 1: few conversations are one-way.
    • a far more interesting measure would be the fraction of conversations that are initiated; which could be construed as desperation when absurdly high.
• these records run only as far back as 17 Jul, which is 1.5 months of data.

• single user, rather rare.
• it was heavily used in July and early August, due to the contact having little else to do.
• that has changed now, and so this is likely to see very little activity.
• hence analytics on such a transient medium are not worth the effort.

• started to be used only around the last week of August.
• have two contacts on telegram, one rare and one with high frequency.
• it's possible to manually count the messsages sent/received on the app/webapp. it's super tedious, but since it's a question of 10 days/being unemployed, I did it.

• asyncronous.
• it's hard to analyze this as I do other IMs.
• I send and receive really long emails, that are composed over the course of at least a week.
• it is hard work writing an email – and nearly impossible for me to do it in one session.
• otoh, it is the most rewarding form of communication – it makes me really happy when I receive a reply or an unsolicited mail from a friend.

• this is the rarest mode of communication.
• it was once exclusively used by a friend; this interaction has now moved to hangouts. (hence its presence)
• I might remove it henceforth if it receives no activity.
  • I could use it to log videochats on hangouts etc. henceforth.

• this happens roughly once every two weeks.
• this number is going to increase.
• I don't keep a good record of it, so haven't included it.
• I intend to keep track of when I last left the house, though.

key:

• columns represent days of the month
• each row represents a person (anonymized)
• the bottom row computes the total number of people contacted that day.
• the final column computes the fraction of days in a month where I am in contact with each person.
• the final cell of the table averages the number of people contacted in a day.
• data for July starts on the 17th.

these use the number of messages as an indication of volume of conversation on hangouts, signal and telegram:

• the first column is the date.
• the next three columns the number of messages sent AND received on α, § and ∇ respectively.
• the fifth column is the net number of messages exchanged across all channels.
• the final column is a sideways manhattan/histogram based on the total volume column.
  • it is generated by orgtbl-ascii-plot or C-c " a
  • the scale of the columns varies from each month, depending on its largest value. The lowest value has to be fixed at 0.
    • $OUTPUT_COL ='(orgtbl-uc-draw-grid $INPUT_COL <min_value> <max_value> <resolution>) is set in #+TBLFM: that follows the table.
    • uc-draw-grid / uc-draw-cont changes the style of the bars to unicode.
  • to reiterate, the heights are relative to each month.
• the last row indicates the average number of messages exchanged per day on each platform.
• I would have preferred a vertical histogram; but orgtbl-ascii-plot is the simplest way to plot data.
  • there did not seem to be an option for a vertical output.
  • transposing the table did not help.
  • gnuplot would have done it, but I am lazy and that's overkill.
  • also, I don't want to deal with inlining images in html etc.