Last modified: 2005-02-12 by rob raeside
Keywords: fotw | bottasini | history |
Links: FOTW homepage | search | disclaimer and copyright | write us | mirrors
by Mark Sensen
|September 1993 -||Giuseppe Bottasini (Italy)||CESI|
|Late 1994 -||Giuseppe Bottasini (Italy)|
|24 July 1997 -||Josh Fruhlinger (USA)||University of California, Berkeley|
|June 1998 -||Rob Raeside (Canada)|
|1 August 1998 -||Edward Mooney (USA)||QNET.com|
|April 2000 -||eGroups.com|
|1 August 2000 -||Ole Andersen (Denmark)|
|January 2001 -||YahooGroups.com|
|1 August 2002||Steve Kramer (USA)|
|1 December 2003||António Martins-Tuválkin (Portugal)|
The mailing list was begun as a discussion group of about a dozen people, including Giuseppe Bottasini, Christopher Vance, and Alessio Bragadini, in about September 1993. It soon expanded to include a regular group of about 100 to 150 addresses. Many of the images initially distributed via the mailing list were written in PostScript by Christopher Vance, who maintained the best flag-site on the Internet at the time. The mailing list was initially managed by Giuseppe Bottasini, an engineer from CESI in Milan, until on July 24 1997 it moved to a majordomo list at University of California, Berkeley, managed by Josh Fruhlinger, an MA candidate in Classical History. On 1 August 1998 the mailing list moved to QNET.com, a majordomo list, managed by Edward Mooney, a social studies teacher in Palmdale, Antelope Valley, California. In April 2000, spam found its way through the majordomo list server, and in spite of heroic efforts by Edward, forced a move from the majordomo to eGroups, which was subsequently acquired by YahooGroups, the current host of the mailing list. In January 2000, Ole Andersen, of Copenhagen, Denmark, was appointed assistant list master, and took over full management of the list on 1 August 2000 for a two-year appointment, and he was replaced by Steve Kramer in 2002. Steve had to relinquish his term in December 2003 due to illness, and was replaced by António Martins-Tuválkin until 1 December 2005. André Coutanche is assistant list master, and the position of apprentice list master is held by Pascal Gross.
Rob Raeside, 27 October 2004
The list was increasingly active in the early days, and has enjoyed about 15,000 to 20,000 messages a year since 1997. Željko Heimer kept track of the number of messages over this time, and reported:
1996 6166 (March-December)
2002 16645 (to end of November)
He added, "Unfortunately, those before I have kept somewhere on some diskette which I don't know where it might be any more. I have joined in mid-1995, but the list exists since September 1993. I don't suppose that the number of messages was larger then that in 1996, probably was even smaller. I suppose we may assume that 1993-1995 did not reach 6000 or so."
Željko Heimer, 1 December 2002
I have taken a look at the number of images sent to the mailing list each year.
|Year||Image files submitted||Total files size||Size/file (kB/image)|
* since August 1995
We may see that the year 2001 set the absolute record in number of files as well as the amount of megabytes and since then we have had a kind of a stagnation - statistically. I leave this for other to explain, but I think that there are still some images (and text) from 2001 that are waiting to be edited.
What is significant, and not obvious in the first glance is the last column that I added after some thinking - it shows the average size of a file image posted to the list. We can notice a steady and almost regular tendency. If it were not for the slight disturbance in 1999 the line would be almost regularly straight, and it almost seems that if someone forgot to send something and 1999 and "caught up" in 2000 (must have been the millennium bug):
Also, it is worth to note that the size of the images has practically doubled from 1995 until now. I also leave this for others to explain it, except for the obvious - since the size of image file depends on its "complexity", it seems that after we finished with the simpler images we slowly took up with the more complicated ones.
Željko Heimer, 18 January 2004
The Flags of the World website was created by Giuseppe Bottasini, of Milan, Italy, in the early days of general public access to the Internet (late 1994). Giuseppe initially operated both the website and mailing list from CESI (Centro Elettrotecnico Sperimentale Italiano), until bandwidth pressures resulting from the site's popularity required him to look for a more permanent site. In January 1997 a devastating system crash at CESI forced the immediate move, briefly through an Israeli mirror, to a site at Digibel in Belgium. Giuseppe continued to manage the website, initially assisted by Željko Heimer, who was responsible for ensuring a flag on every page, and who standardized the layout of the pages, the image sizes and palettes and introduced the use of ISO codes for file names. Željko "retired" when he went to do military service in Croatia in late 1997, although he has intermittently remained active as an editor. Rob Raeside took over as assistant, until in June 1998 Giuseppe announced his resignation as director and Rob took over. The website continued to grow, exceeding 100 Mbytes in size in October 2000. In May 2001, the "home site" at Digibel in Belgium closed, and FOTW continued as a dispersed series of mirrors operating in USA, Switzerland, United Kingdom, Australia, Germany, Puerto Rico, and Russia.
by Giuseppe Bottasini
|Year||Pages||Images||Pages + Images||Images/page||Yearly growth*||Absolute Growth*||Source|
* Using 1997 as a base.
Giuseppe Bottasini, 09 August 2004
The data are very interesting and give some clue on the growth of our website. The curves for both numbers of pages and images follow an exponential pattern. An exponential growth can be expected in a biological system when there is no resource limitation, for example when a bacterial population is allowed to grow without space limitation or when a disease epidemics is allowed to spread over an infinite number of susceptible hosts. Excluding the 2003 data, I fitted a simple exponential model to the remaining data, using a linear exponential function and a least square regression method. To be simpler, linearization transformed the original curves into approximated straight lines and regression yielded the straight line which best fit the data. As expected, the goodness-of-fit is high and the exponential model takes
into account 96% of the data variability for pages and 97% for images. However, the goodness-of-fit is exaggerated because of the small size of the data set.
Using the regression equations, it is possible to make forecasts of the future growth of our website. This is arithmetically easy but statistically hazardous. One of the first conclusion learned in statistics classes is that a model is not necessarily valid outside its original domain of validation. This is a reality several economists and
politicians do not want to hear. Extending my model to year 2003, I would predict 33,472 pages and 67,541 images for the end of the year, whereas Guiseppe's census for mid-2003 was only 19,000 pages and 36,000 images. The big discrepancy between the probable size of our website at the end of 2003 and my prediction clearly indicates that the exponential model is no longer suitable for explaining the growth of our website. In fact, the exponential model is not suitable for most biological systems because if it based on the very strong assumption of unlimited resource. Here, this would mean that the number of flags to be found is infinite. Although there will always be flags we don't show yet, this number shall progressively decrease due to our ferreting, reporting and editing effort. It seems from the data that the growth curve of our website begins to level off, and the most probable shape of the curve is a sigmoid, i.e. an S-shaped curve. The first part of the S corresponds to years 1997-1198 (and probably the earlier years), during which the overall dynamics of our website was fairly slow. Then the dynamics turned to an epidemic process with a nearly constant increase year per rate. We are probably now near the inflexion point of the curve.
To conclude, the overall pattern of the growth curve of pages and images shows that the dynamics of our website has been following a regular and optimal pattern for the last years. Of course, the basic model I have used does not encompass some of the specificities of our dynamics, such as the appearance of new flags (limited renewable resource) and the process of splitting pages when they become too big.
Ivan Sache, 23 August 2003