What Effect Does the Orphaning Process Have Upon Games?

Mark Nelson

(Reprinted from Everything 92, June 1995)


In an earlier article, I analyzed completed regular Diplomacy games reported in Everything #85 (May 1992) through Everything #88 (September 1993). The games were divided into games played postally and games played over the CompuServe network. In order to compare postal games to CompuServe games, I used three indicator functions: the percentage number of games finishing in a win (the WIN number), the average number of dropouts per game (the DROPOUT number) and the average length of a game (the LENGTH number). The dependence of the WIN number upon the DROPOUT number was also examined.

By an "orphan" game I mean either a game which moved from one 'zine to another, usually with a change in GM, or which had a change in GM while remaining in the same 'zine. In the earlier article, orphaned games were discarded from my datasets because "these processes often cause dropouts." The purpose of the present article is to compare indicator functions for non-orphaned games to orphaned games in order to determine what effect, if any, the orphaning process has on a game. This article considers only postal games because the number of orphaned games on CompuServe (6) is too small.

Glancing through back issues of Everything, orphan games appear to belong to one of two distinct categories. Frequently, a couple of players will drop out in the period in which the game is being orphaned and when the game restarts the remaining players from the original 'zine will either agree a draw or a concession. Alternatively, new players may enter the game and their enthusiasm ensures that the game continues for many more years. When games finish shortly after restarting, it is natural to ask if this result is the "right" result. Does orphaning affect the game result? Does orphaning affect how long the game will last?

To examine these questions we will use a naive approach: a game is either orphaned or not orphaned. A more sophisticated approach would try to measure the "completion index" of a game at the time of the fold. By "completion index," I mean a measure of how near the game has come to finishing. The variation of the indicator functions with completion index would then be analyzed. A more sophisticated approach might also examine the effect of multiple orphaning and, perhaps most importantly, any effect that the time-delay between a 'zine folding and ita games restarting might have on the indicator functions. Do "smooth" folds have less effect on games than "poor" folds?

Games reported in Everything #85 (May 1992) through Everything #91 (March 1995) are considered. During this period there were 181 non-orphaned games reported and 67 orphaned games. Of the orphaned games, two were games in which there was only a change in the GM. I have discarded these games from the survey.

Statistical Analysis

Of the 247 regular postal games reported in this three-year period, 65 (26.4%) of them were orphaned. In other words, one in four reported games were orphaned. Is this a historically low, average, or high value? Is it an acceptable rate? By examining back-issues of Everything, I hope to answer the former question at a later date.

Table 1 shows what values the indicator functions take on the two datasets. I am very surprised to discover that the average year of finish between non-orphaned and orphaned games is indistinguishable. This in itself doesn't mean much. It could be a fluke, in the sense that orphan games are on the whole either much shorter or considerable longer than non-orphaned games and the figures average to the non-orphaned value.

Table 1. Indicator Functions on Non-Orphaned and Orphaned Postal Games
Number of Games18165
Games Won8742
Games Drawn9423
Percents of Games Won48.07%64.62%
Average Game-Year of Finish1910.42 +/- 4.14910.32 +/- 2.90
Average Dropouts per game2.75 +/- 1.564.00 +/- 1.32

Table 2 shows the percentage of games finish in a given year or spread of years. Although the sample size is small, the the standard deviation for non-orphan and orphaned games are similar, and it appears that the orphaning process has no effect on the game-length.

Table 2. Percentage of Games Finishing in Given Game-Year(s)
Non-OrphanedOrphanedNon-Orphaned Orphaned
19030.60.019128.3 6.3
19043.33.119139.9 4.7
19053.30.019145.5 7.8
19066.13.119151.1 7.8
19089.921.91921-19251.7 0.0
190912.27.81926-19300.0 0.0
191013.812.51931-19350.0 0.0
191111.617.91936-19400.6 0.0

As expected, and as predicted in the earlier article, orphaned games have more dropouts than non-orphaned games. On average, there are 1.25 more dropouts in orphaned games than in non-orphaned games.

The tendency for more dropouts to occur in orphaned games is also demonstrated in Table 3, below: nearly 50% of all non-orphaned games have two or fewer, dropouts, whilst the corresponding figure for orphaned games is only 17%. Finally, observe that the WIN number for orphaned games is approximately 125% greater than the corresponding number for non-orphaned games -- orphaned games are more likely to finish in a win.

Table 3: Percent of Games Finishing With A Given Number of Dropouts
Number of Dropouts
Non-orphaned6.6%17.1%22.7%20.4% 17.7%12.2%3.3%0.0%

When non-orphaned games are analyzed, there is a (non-linear) relationship between the number of dropouts and the chance that the game finishes in a win. This is most easily seen when the data is presented as a rolling average. Is the same phenomenon observed in orphaned games? Are orphaned games with few dropouts more, or less, akin to their non-orphaned counterparts than those games with many dropouts? Table 4 provides some inconclusive answers.

Table 4. Percentage of Games Finishing in a Win as a Function of the Rolling Number of Dropouts
0 to 244.1%63.6%
1 to 343.1%70.8%
2 to 450.0%66.7%
3 to 553.9%63.0%
4 to 656.7%60.0%
5 to 753.4%61.5%

Regardless of the number of dropouts, orphaned games are more likely to end in a win than their non-orphaned counterpart. Surprisingly, this trend is more distinctive for games with fewer dropouts than for games with more dropouts. This is a tentative result, since the survey sample for orphaned games with few dropouts is small.

We have noticed that orphaning affects the WIN number. Does it also affect the type of draw? To answer this question I have devised the DRAW SPECTRUM. The DRAW SPECTRUM is the percentage number of drawn-games which finish in a n-way draw. Before examining the data presented in Table 5, recall that the sample size of drawn orphaned games is very small (22 games).

Table 5. The DRAW SPECTRUM for Non-Orphaned and Orphaned Games
Draw Participants
23456 7
Non-Orphaned43.0%28.0%19.4%7.5% 2.2%0.0%
Orphaned27.3%40.9%31.8%0.0% 0.0%0.0%


In this article, non-orphaned and orphaned postal games reported in Everything $85 (May 1992) through Everything #91 (March 1995) have been analyzed, with the following results:


  1. When the change in the name of the zine was considered to be "cosmetic," games in that 'zine were not treated as orphans. A cosmetic change means that either that the editor changed the name of the 'zine, but nothing else changed, or that the editor folded the 'zine but continued to run games by flyer without a noticeable break.
  2. The following name changes were considered to be cosmetic: The Boob Report/The Abyssinian Prince, Cathy's Ramblings/Rambling By Moonlight, Dipadeedoodah!/flyer/Akrasia, Penguin Dip/Black Tie Affairs, Vertigo/Meet George Jetson and Don Del Grande's various 'zines.

Back to the article list