import re
import csv
from collections import Counter

import numpy as np
import requests
import matplotlib.pyplot as plt
import seaborn

request = requests.get('https://raw.githubusercontent.com/TheSCInitiative/BOINC/main/BOINC_Census/2023/data_source/BOINC_Census_2023_results_raw.csv')
request.encoding = request.apparent_encoding
data_str = request.text
data = [row for row in csv.reader(data_str.split('\n')) if bool(row)]

# Remove headers from data
headers = data[0]
data = data[1:]

# Eye candy settings
BINARY_COLOURS = ['#38CF62', '#E65050']
GENERAL_COLOURS = seaborn.color_palette("colorblind")
BOINC_BLUE = '#163E72'

def get_col(heading: str, data_source: list):
    idx = headers.index(heading)
    return [row[idx] for row in data_source]

def multi_sel_to_hist(mult_sel_col: list[str]):
    '''
    Converts a column with CSV values in each cell to a simple list that can
    be turned into a histogram.
    
    Fillout does not perform escaping on internal commas so we have to handle it.
    Because I always put a space after a comma, we can safely replace it.
    '''
    return sum([row.replace(', ', ' ').split(',') for row in mult_sel_col], start=[])

def calc_other_from_multi_select(data: list[tuple[str, int]], base_responses: list[str]) -> int:
    other_sum = 0

    for response, count in data:
        if response not in base_responses:
            other_sum += count

    return other_sum

print(f'Completed Submissions: {len(data)}')

fig, ax = plt.subplots(1, 2)
fig.set_figwidth(15)

used_boinc_counter = Counter(get_col('Do you or have you ever used BOINC?', data))
completed_census_counter = Counter(get_col('Have you completed a previous BOINC Census before?', data))

ax[0].pie([used_boinc_counter['Yes'], used_boinc_counter['No']], labels=['Yes', 'No'], autopct='%.0f%%', colors=BINARY_COLOURS)
ax[1].pie([completed_census_counter['Yes'], completed_census_counter['No']], labels=['Yes', 'No'], autopct='%.0f%%', colors=BINARY_COLOURS)
ax[0].title.set_text('Do you or have you ever used BOINC?')
ax[1].title.set_text('Have you completed a previous BOINC Census before?')
plt.show()

Completed Submissions: 552

fig, ax = plt.subplots(1, 2)
fig.set_figwidth(15)
gender_groups = Counter(get_col('What gender do you identify as?', data)).items()
age_groups = Counter(get_col('What is your age?', data))
age_groups_sorted = sorted(age_groups.items(), key=lambda v: ord(v[0][0]) + 999 * int(v[0].endswith('+'))) # Ensure that 'more than' values are put to the end

ax[0].bar([a[0] for a in gender_groups], [a[1] for a in gender_groups], color=BOINC_BLUE)
ax[1].bar([a[0] for a in age_groups_sorted], [a[1] for a in age_groups_sorted], color=BOINC_BLUE)
ax[0].title.set_text('What gender do you identify as?')
ax[1].title.set_text('What is your age?')

plt.show()

fig, ax = plt.subplots(1, 2)
fig.set_figwidth(15)
edu_groups = Counter(get_col('What is your highest level of education you have completed?', data))
country_groups = Counter(get_col('What country are you from?', data))
edu_groups = sorted(edu_groups.items(), key=lambda x: x[1], reverse=True)
# Take the top 5 sources to prevent clutter
country_groups = sorted(country_groups.items(), key=lambda x: x[1], reverse=True)[:5]

repl_dict = {
    'United States of America' : 'US',
    'United Kingdom' : 'UK',
    'Tertiary Education (postgraduate)' : 'Uni Postgrad', 
    'Tertiary Education (undergraduate university or other institution)' : 'Uni Undergrad', 
    'Secondary Education (high school)' : 'Secondary', 
    'Primary Education (kindergarten or primary school)' : 'Primary'
}

ax[0].bar([repl_dict.get(a[0], a[0]) for a in edu_groups if a[0] in repl_dict], [a[1] for a in edu_groups if a[0] in repl_dict], color=BOINC_BLUE)
ax[1].bar([repl_dict.get(a[0], a[0]) for a in country_groups], [a[1] for a in country_groups], color=BOINC_BLUE)
ax[0].title.set_text('What is your highest level of education?')
ax[1].title.set_text('What country are you from? (top-five)')

plt.show()

fig, ax = plt.subplots(1, 3)
fig.set_figwidth(15)
computers = Counter(get_col('How many computers (including mobile devices) do you have that run BOINC regularly?', data))
computers_sorted = sorted(computers.items(), key=lambda v: ord(v[0][0]) + 999 * int(v[0].endswith('+'))) # Ensure that 'more than' values are put to the end
mobiles = Counter(get_col('How many of the above computers are mobile devices? (mobile phones or tablets)', data))
mobiles_sorted = sorted(mobiles.items(), key=lambda v: ord(v[0][0]) + 999 * int(v[0].endswith('+'))) # Ensure that 'more than' values are put to the end

ax[0].bar([c[0] for c in computers_sorted], [c[1] for c in computers_sorted], color=BOINC_BLUE)
ax[1].bar([c[0] for c in mobiles_sorted], [c[1] for c in mobiles_sorted], color=BOINC_BLUE)
ax[2].pie([len(data) - mobiles['0'], mobiles['0']], labels=['BOINC on mobile', 'No BOINC mobile'], autopct='%.0f%%', colors=BINARY_COLOURS)
ax[0].title.set_text('How many devices do you have running BOINC?')
ax[1].title.set_text('How many are mobile devices?')
ax[2].title.set_text('BOINC on mobile rate')

plt.show()

# boinc_often = Counter(get_col('How often do you run BOINC?', data))
# boinc_often = sorted(boinc_often.items(), key=lambda x: x[1], reverse=True)

# repl_dict = {
#     'A few days of the week' : 'Few days/w',
#     'All the time (24/7)' : '24/7',
#     'Most days of the week' : 'Most days', 
#     'A few hours every week' : '<1day/w', 
#     "I don't use BOINC" : 'N/A', 
#     "Only when I'm not using the computer" : 'Only Idle'
# }

# plt.bar([repl_dict.get(a[0], a[0]) for a in boinc_often], [a[1] for a in boinc_often], color=BOINC_BLUE)
# plt.title('How often do you run BOINC?')
# plt.show()

fig, ax = plt.subplots(1, 2)
fig.set_figwidth(15)

num_boinc_projects = Counter(get_col('How many BOINC projects are you signed up to?', data))
num_boinc_projects_sorted = sorted(num_boinc_projects.items(), key=lambda v: ord(v[0][0]) + 999 * int(v[0].endswith('+'))) # Ensure that 'more than' values are put to the end
regular_cruncher_num_boinc_projects = Counter(get_col('Out of those BOINC projects, how many do you regularly crunch for?', data))
regular_cruncher_num_boinc_projects_sorted = sorted(regular_cruncher_num_boinc_projects.items(), key=lambda v: ord(v[0][0]) + 999 * int(v[0].endswith('+'))) # Ensure that 'more than' values are put to the end

ax[0].bar([c[0] for c in num_boinc_projects_sorted], [c[1] for c in num_boinc_projects_sorted], color=BOINC_BLUE)
ax[1].bar([c[0] for c in regular_cruncher_num_boinc_projects_sorted], [c[1] for c in regular_cruncher_num_boinc_projects_sorted], color=BOINC_BLUE)
ax[0].title.set_text('How many BOINC projects are you signed up to?')
ax[1].title.set_text('Out of those BOINC projects, how many do you regularly crunch for?')

plt.show()

fig, ax = plt.subplots(1, 3)
fig.set_figwidth(15)

has_competed = Counter(get_col('Have you ever competed against someone else to achieve a higher rank on BOINC leaderboards?', data))
joined_team = Counter(get_col('Have you joined a BOINC team?', data))
use_boinc_from_team = Counter(get_col('Do you use BOINC more because you are competing in a team?', data))


ax[0].pie([has_competed['Yes'], has_competed['No']], labels=['Yes', 'No'], autopct='%.0f%%', colors=BINARY_COLOURS)
ax[1].pie([joined_team['Yes'], joined_team['No']], labels=['Yes', 'No'], autopct='%.0f%%', colors=BINARY_COLOURS)
ax[2].pie([use_boinc_from_team['Yes'], use_boinc_from_team['No or Not sure']], labels=['Yes', 'No or Not sure'], autopct='%.0f%%', colors=BINARY_COLOURS)
ax[0].title.set_text('Have you ever competed against someone\non the BOINC leaderboards?')
ax[1].title.set_text('Have you joined a BOINC team?')
ax[2].title.set_text('Do you use BOINC more because you are in a team?')

plt.show()

fig, ax = plt.subplots(1, 2)
fig.set_figwidth(15)

global_impact = Counter(get_col('How much impact do you believe BOINC is having on the world?', data)).items()
global_impact_sorted = sorted(global_impact, key=lambda v: ord(v[0][0]) + 999 * int(v[0].endswith('+')))
contrib_impact = Counter(get_col('How much impact do you think your own BOINC contributions are making?', data)).items()
contrib_impact_sorted = sorted(contrib_impact, key=lambda v: ord(v[0][0]) + 999 * int(v[0].endswith('+')))

ax[0].bar([a[0] for a in global_impact_sorted], [a[1] for a in global_impact_sorted], color=BOINC_BLUE)
ax[1].bar([a[0] for a in contrib_impact_sorted], [a[1] for a in contrib_impact_sorted], color=BOINC_BLUE)
ax[0].title.set_text('How much impact do you believe BOINC is having on the world?')
ax[1].title.set_text('How much impact do you think your own BOINC contributions are making?')

plt.show()

fig, ax = plt.subplots(1, 1)
fig.set_figwidth(15)

contributions = list(Counter(multi_sel_to_hist(get_col('Aside from running BOINC, how else have you contributed to BOINC or computing projects running on BOINC?', data))).items())

repl_dict = {
    'No additional contribution' : 'N/A',
    'Made a financial donation' : 'Financial donation',
    'Contributed my time or did volunteer work' : 'Volunteer work',
    'Made a non-financial donation (eg donated server hardware)' : 'Non-financial donation',
    'Other' : 'Other'
}
num_other = calc_other_from_multi_select(contributions, list(repl_dict.keys()))
contributions.append(('Other', num_other))

ax.pie([a[1] for a in contributions if a[0] in repl_dict], labels=[repl_dict.get(a[0], a[0]) for a in contributions if a[0] in repl_dict], autopct='%.0f%%', colors=GENERAL_COLOURS)
ax.title.set_text('How else have you contributed to BOINC?')

plt.show()

fig, ax = plt.subplots(1, 2)
fig.set_figwidth(15)
when_boinc = list(sorted(Counter(get_col('What year did you first hear about BOINC?', data)).items()))
where_boinc = Counter(get_col('How did you first hear about BOINC?', data))
# Take the top 5 sources to prevent clutter
where_boinc_top = sorted(where_boinc.items(), key=lambda x: x[1], reverse=True)[:5]

repl_dict = {
    'Via an internet search' : 'Web search',
    'From a news article (both print or digital)' : 'News',
    'From social media' : 'Social media',
    '2002-2010' : "'02-'10",
    '2011-2018' : "'11-'18",
    'Before 2002' : '<2002'
}

# Workaround for "Before 2002"
when_boinc.insert(0, when_boinc.pop(-1))

ax[0].bar([repl_dict.get(a[0], a[0]) for a in when_boinc], [a[1] for a in when_boinc], color=BOINC_BLUE)
ax[1].bar([repl_dict.get(a[0], a[0]) for a in where_boinc_top], [a[1] for a in where_boinc_top], color=BOINC_BLUE)
ax[0].title.set_text('What year did you first hear about BOINC?')
ax[1].title.set_text('How did you first hear about BOINC? (top-five)')

plt.show()

fig, ax = plt.subplots(1, 1)
fig.set_figwidth(15)

social_media = list(Counter(multi_sel_to_hist(get_col('On what social media platforms do you want to hear about BOINC?', data))).items())

responses = {'Reddit', 'Twitter/X', 'Facebook', 'Mastodon', 'Instagram', 'Linkedin', 'Lemmy', 'Other'}
num_other = calc_other_from_multi_select(social_media, responses)
social_media.append(('Other', num_other))

social_media_sorted = sorted(social_media, key=lambda x: x[1], reverse=True)

ax.bar([a[0] for a in social_media_sorted if a[0] in responses], [a[1] for a in social_media_sorted if a[0] in responses], color=BOINC_BLUE)
ax.title.set_text('On what social media platforms do you want to hear about BOINC?')

plt.show()

# fig, ax = plt.subplots(1, 2)
# fig.set_figwidth(15)
# where_help = Counter(multi_sel_to_hist(get_col('Where do you seek help for problems you have with running BOINC?', data)))
# where_hear = Counter(multi_sel_to_hist(get_col('In what ways would you like to hear about what BOINC projects are doing?', data)))
# # Take the top 5 options to prevent clutter
# where_help_top = sorted(where_help.items(), key=lambda x: x[1], reverse=True)[:5]
# where_hear_top = sorted(where_hear.items(), key=lambda x: x[1], reverse=True)[:5]

# repl_dict = {
#     'BOINC project forums/message boards' : 'Proj. forums',
#     'Google search or other search engine' : 'Web search',
#     'The official BOINC forum/message board' : 'BOINC forum',
#     'The BOINC website' : 'BOINC website',
#     'Chat groups like Discord or Telegram' : 'Chat groups',
#     'Via the BOINC manager' : 'BOINC manager',
#     'Via a web page or dashboard' : 'Web page',
#     'Email notifications' : 'Email',
#     'Forum/message board posts' : 'Forum post'
# }

# ax[0].bar([repl_dict.get(a[0], a[0]) for a in where_help_top], [a[1] for a in where_help_top], color=BOINC_BLUE)
# ax[1].bar([repl_dict.get(a[0], a[0]) for a in where_hear_top], [a[1] for a in where_hear_top], color=BOINC_BLUE)
# ax[0].title.set_text('Where do you seek help for BOINC? (top-five)')
# ax[1].title.set_text('How do you want to hear about BOINC? (top-five)')

# plt.show()

fig, ax = plt.subplots(1, 2)
fig.set_figwidth(15)
aspects = Counter(multi_sel_to_hist(get_col('What are the most important aspects of a BOINC project?', data)))
wu_timings = Counter(multi_sel_to_hist(get_col('How long do you prefer your work units to run before finishing?', data)))
# Take the top 5 options to prevent clutter
aspects_top = sorted(aspects.items(), key=lambda x: x[1], reverse=True)[:5]
wu_timings_top = sorted(wu_timings.items(), key=lambda x: x[1], reverse=True)[:5]

repl_dict = {
    'The work they do is beneficial to the world' : 'Work impact',
    'Has CPU work units' : 'CPU Work',
    'Has GPU work units' : 'GPU Work',
    'Regularly publishes results/academic papers' : 'Freq. results',
    'Active project staff' : 'Active staff',
    'Up to 1 hour' : '<1hr',
    'Up to 30 minutes' : '<30min',
    'Only a few minutes' : 'A few minutes',
}

ax[0].bar([repl_dict.get(a[0], a[0]) for a in aspects_top], [a[1] for a in aspects_top], color=BOINC_BLUE)
ax[1].bar([repl_dict.get(a[0], a[0]) for a in wu_timings_top], [a[1] for a in wu_timings_top], color=BOINC_BLUE)
ax[0].title.set_text('What are the important aspects of a project? (top-five)')
ax[1].title.set_text('How long do you prefer your work units to run before finishing? (top-five)')

plt.show()

fig, ax = plt.subplots(1, 2)
fig.set_figwidth(15)
boinc_proj = list(Counter(get_col('Have you ever created a BOINC project or wanted to create one?', data)).items())
boinc_proj.sort(reverse=True)
boinc_diff = Counter(multi_sel_to_hist(get_col('What are the biggest difficulties in making a BOINC project?', data)))
# This was a hidden question so we got a lot of empty results
boinc_diff.pop('')
# Take the top 5 sources to prevent clutter
boinc_diff = sorted(boinc_diff.items(), key=lambda x: x[1], reverse=True)[:5]
repl_dict = {
    'Technical skills' : 'Tech skills',
    'No or inadequate funding' : 'Money',
    'Time commitments' : 'Time',
    'Server hosting/Computer hardware' : 'Hosting',
    'My idea is too small to be worth the effort' : 'Idea too small',
    'Inadequate BOINC documentation' : 'Docs'
}

ax[0].pie([a[1] for a in boinc_proj], labels=[a[0] for a in boinc_proj], autopct='%.0f%%', colors=BINARY_COLOURS)
ax[1].bar([repl_dict.get(a[0], a[0]) for a in boinc_diff], [a[1] for a in boinc_diff], color=BOINC_BLUE)
ax[0].title.set_text('Have you ever wanted to start a BOINC project?')
ax[1].title.set_text('What are the biggest difficulties? (top-five)')

plt.show()

fig, ax = plt.subplots(1, 1)
fig.set_figwidth(15)

compensation = Counter(get_col('Would you use BOINC more if you were paid a small amount of money for crunching?', data)).items()

ax.pie([a[1] for a in compensation], labels=[a[0] for a in compensation], autopct='%.0f%%', colors=BINARY_COLOURS)
ax.title.set_text('Would you use BOINC more if you were paid\na small amount of money for crunching?')

plt.show()

fig, ax = plt.subplots(1, 1)

other_dcn = Counter(multi_sel_to_hist(get_col('Do you use any other distributed computing platforms?', data))).items()
other_dcn_sorted = sorted(other_dcn, key=lambda x: x[1], reverse=True)

responses = {'No', 'Folding@home', 'GIMPS', 'Dreamlabs', 'distributed.net'}

ax.bar([a[0] for a in other_dcn_sorted if a[0] in responses], [a[1] for a in other_dcn_sorted if a[0] in responses], color=BOINC_BLUE)
ax.title.set_text('Do you use any other distributed computing platforms?')

plt.show()

Results of the BOINC Census 2023¶

Introduction¶

Basic Statistics¶

Demographics¶

BOINC usage¶

BOINC Impact¶

BOINC Origins¶

Project Administration¶

Miscellaneous¶

Your feedback¶

Conclusion¶

Summary¶

Improvements for next Census¶

Closing remarks¶

Get ready for the next Census in November 2024!¶

Links¶