First-time maintainers
How many people publish a first R package each year, and how many of them are still publishing years later. The names behind a year are in that year's digest.
- 46,072
- People tracked
- 26
- Arrival years
- 48%
- Still at 5 yr (2019)
3,006
people published a first package in 2026
People whose earliest package appeared that year, counted once.
Whether they stay
Each row is the people who arrived that year. Reading across shows how many were still publishing at each anniversary; reading down a column compares one cohort against the next.
| Arrived | People | 1 yr | 2 yr | 3 yr | 4 yr | 5 yr |
|---|---|---|---|---|---|---|
| 1999 | 1 | 0.0%, 0 of 1 | 0.0%, 0 of 1 | 0.0%, 0 of 1 | 0.0%, 0 of 1 | 0.0%, 0 of 1 |
| 2000 | 55 | 90.9%, 50 of 55 | 90.9%, 50 of 55 | 90.9%, 50 of 55 | 89.1%, 49 of 55 | 83.6%, 46 of 55 |
| 2001 | 61 | 75.4%, 46 of 61 | 75.4%, 46 of 61 | 72.1%, 44 of 61 | 70.5%, 43 of 61 | 68.9%, 42 of 61 |
| 2002 | 88 | 85.2%, 75 of 88 | 81.8%, 72 of 88 | 76.1%, 67 of 88 | 70.5%, 62 of 88 | 65.9%, 58 of 88 |
| 2003 | 111 | 87.4%, 97 of 111 | 82.0%, 91 of 111 | 77.5%, 86 of 111 | 74.8%, 83 of 111 | 73.0%, 81 of 111 |
| 2004 | 206 | 91.7%, 189 of 206 | 76.7%, 158 of 206 | 73.3%, 151 of 206 | 67.5%, 139 of 206 | 63.6%, 131 of 206 |
| 2005 | 304 | 86.8%, 264 of 304 | 82.9%, 252 of 304 | 75.3%, 229 of 304 | 73.0%, 222 of 304 | 69.4%, 211 of 304 |
| 2006 | 443 | 87.8%, 389 of 443 | 82.2%, 364 of 443 | 78.1%, 346 of 443 | 75.6%, 335 of 443 | 65.7%, 291 of 443 |
| 2007 | 485 | 80.6%, 391 of 485 | 75.5%, 366 of 485 | 72.2%, 350 of 485 | 67.6%, 328 of 485 | 63.5%, 308 of 485 |
| 2008 | 575 | 84.7%, 487 of 575 | 76.5%, 440 of 575 | 73.0%, 420 of 575 | 68.3%, 393 of 575 | 62.4%, 359 of 575 |
| 2009 | 792 | 82.1%, 650 of 792 | 74.1%, 587 of 792 | 68.7%, 544 of 792 | 63.8%, 505 of 792 | 59.3%, 470 of 792 |
| 2010 | 960 | 81.0%, 778 of 960 | 75.0%, 720 of 960 | 69.5%, 667 of 960 | 62.7%, 602 of 960 | 58.5%, 562 of 960 |
| 2011 | 1,226 | 78.5%, 963 of 1,226 | 70.1%, 860 of 1,226 | 64.7%, 793 of 1,226 | 59.6%, 731 of 1,226 | 55.3%, 678 of 1,226 |
| 2012 | 1,607 | 79.1%, 1,271 of 1,607 | 71.8%, 1,154 of 1,607 | 65.3%, 1,049 of 1,607 | 62.3%, 1,001 of 1,607 | 58.5%, 940 of 1,607 |
| 2013 | 1,902 | 76.3%, 1,452 of 1,902 | 69.1%, 1,315 of 1,902 | 64.1%, 1,220 of 1,902 | 60.7%, 1,155 of 1,902 | 57.4%, 1,092 of 1,902 |
| 2014 | 2,144 | 78.4%, 1,680 of 2,144 | 70.0%, 1,501 of 2,144 | 64.8%, 1,389 of 2,144 | 61.9%, 1,327 of 2,144 | 58.1%, 1,246 of 2,144 |
| 2015 | 2,835 | 78.1%, 2,214 of 2,835 | 69.6%, 1,974 of 2,835 | 65.0%, 1,844 of 2,835 | 61.2%, 1,736 of 2,835 | 57.3%, 1,625 of 2,835 |
| 2016 | 3,213 | 75.2%, 2,416 of 3,213 | 67.6%, 2,171 of 3,213 | 61.3%, 1,971 of 3,213 | 57.3%, 1,842 of 3,213 | 52.9%, 1,700 of 3,213 |
| 2017 | 3,758 | 75.5%, 2,838 of 3,758 | 66.6%, 2,504 of 3,758 | 61.4%, 2,308 of 3,758 | 55.4%, 2,083 of 3,758 | 51.3%, 1,926 of 3,758 |
| 2018 | 3,590 | 77.1%, 2,767 of 3,590 | 68.8%, 2,470 of 3,590 | 61.6%, 2,212 of 3,590 | 56.0%, 2,012 of 3,590 | 50.4%, 1,809 of 3,590 |
| 2019 | 3,474 | 78.4%, 2,722 of 3,474 | 69.4%, 2,412 of 3,474 | 62.7%, 2,178 of 3,474 | 56.0%, 1,945 of 3,474 | 48.0%, 1,668 of 3,474 |
| 2020 | 4,067 | 78.3%, 3,186 of 4,067 | 70.2%, 2,855 of 4,067 | 63.0%, 2,563 of 4,067 | 53.8%, 2,190 of 4,067 | 45.8%, 1,862 of 4,067, provisional |
| 2021 | 3,930 | 74.3%, 2,920 of 3,930 | 64.1%, 2,518 of 3,930 | 55.4%, 2,179 of 3,930 | 44.6%, 1,753 of 3,930, provisional | not yet reached– |
| 2022 | 3,415 | 71.6%, 2,445 of 3,415 | 58.5%, 1,999 of 3,415 | 48.1%, 1,643 of 3,415, provisional | not yet reached– | not yet reached– |
| 2023 | 3,326 | 68.1%, 2,266 of 3,326 | 53.3%, 1,772 of 3,326, provisional | not yet reached– | not yet reached– | not yet reached– |
| 2024 | 3,504 | 62.4%, 2,185 of 3,504, provisional | not yet reached– | not yet reached– | not yet reached– | not yet reached– |
Stronger colour is a larger share still publishing. An en dash means the cohort has not reached that anniversary. A hatched figure is provisional: fewer than two complete years have been observed past it.
What this counts
A first-time maintainer is someone whose earliest package appeared that year. They are counted once, in that year only, however many packages they go on to publish.
Still publishing is not the same as still maintaining. The underlying record is the date someone last appeared in a package's DESCRIPTION, which is a release date. Somebody whose package is live, stable and needs no changes stops counting here, even though nothing has gone wrong. The data cannot tell that apart from leaving, so this page does not claim to.
Why recent years read low
The figures are right-censored. Someone who arrived in 2022 and publishes again in 2027 counts as departed today, while an older cohort has had far longer to come back. An anniversary falling in the current year is withheld entirely rather than published against a part-finished year.
Where a person can split in two
Maintainers are keyed by name as it appears in DESCRIPTION, so a person who changes how they are credited reads as a departure and a fresh arrival. That inflates the newcomer count and depresses retention.