Order Without Office: Whether Standing on Informal Games Ladders Carries Into Wider Regard, from 842 Players in 14 Ladder Books and 670 Non-Players on Chiark Orbital, 1790–1798 CE

1. Introduction

The Culture plays games with a devotion that visitors from other civilisations often find hard to place. Citizens play on Orbitals and aboard ships, alone and in groups, for an evening or for decades, at games of every complexity that Minds and people have devised. Wherever a game is played often enough, its players tend to keep a ladder: a list of names in order of demonstrated skill, held by a club or by a few enthusiasts and rearranged by challenge. A ladder has no authority behind it. Nobody is obliged to appear on one, to accept a challenge or to stay once listed, and no office exists that could enforce a placing.

That absence gives the ladder an unusual interest. The Culture has no wage labour, no ownership of the means of living and no formal hierarchy. Reputation, the regard of other people, is what a citizen's standing consists of, and it is earned and lost by the day (Wescombe, 1712 CE). A ladder is a public ordering of people by one kind of excellence, set up inside a society that otherwise declines to order people at all. Two readings are open. We take two readings as our hypotheses. On the first, the ladder is sealed inside the game: a high place is a pleasure in the play and means nothing outside it. On the second, the order leaks, and a person who stands high in a game is, to some degree, held in higher regard everywhere, as the regard commonly given to outstanding players might lead one to expect.

Contact's field reports record that in some other civilisations skill in games and standing in society are bound closely together (Cadrenne, 1789 CE). Whether anything of the kind happens in the Culture has not been measured. Earlier descriptive work established that informal ladders exist on several Orbitals and set out how challenges are made and accepted (Ildreth, 1745 CE), and that standing in games played without stakes is stable over years (Torvane, 1768 CE). Neither work asked what a ladder place is worth beyond the game.

This paper is written in 1800 CE and uses ladder books kept on Chiark Orbital between 1790 and 1797 CE and a survey of players and non-players carried out there in 1798 CE. We ask three questions. Do people who play ladder games report more regard outside play than those who do not? Within a ladder, does higher standing go with more regard outside play? And is any such gradient shallow or steep, and is it steeper at the top of a ladder than elsewhere?

2. Methods

Chiark Orbital has a long tradition of club play. With the agreement of the clubs concerned, we collected the ladder books that clubs had chosen to lodge with Chiark Hub between 1790 and 1797 CE (Chiark Orbital Hub Records, 1790–1797 CE). The Hub holds these copies as a courtesy and takes no part in keeping them. We call the collection the ladder-book series. It, the survey described below and every measure derived from them are our own compilation for this study, and no Hub record of them existed before. The series holds 14 books, nine for table games and five for physical sports, with 1,068 named entries in all. Each book gives its entries in order of place at the close of the 1797 CE season. Many entries are handles or nicknames, so we asked each club's organiser to identify the resident behind each one. Entries that nobody could place were left out: 52 of the 1,068 (4.9%). The other 1,016 were merged into 979 distinct residents, 37 of whom appeared in two books, and each resident was assigned to the book of the game they reported playing most. All later counts are of these 979 people. To check the identification, two of us independently coded a stratified sample of 200 entries, 100 drawn from entries the organisers had marked uncertain and 100 from the rest, as resolved, unresolved or ambiguous. Agreement was 87.5% and Cohen's kappa 0.79. The sample over-represents hard cases, so these figures are a conservative indicator and do not describe the series as a whole. Disagreements and ambiguous entries were settled by discussion with the organiser, and an entry still ambiguous after that was counted as unplaced, which gives the split of placed and unplaced entries above.

In 1798 CE we invited all 979 resolved players to complete a short survey, by message from the authors; Chiark Hub passed the messages on but did not see replies, in keeping with the custom that Hubs do not read their residents. For comparison we drew 1,200 residents at random from the Hub's resident index for 1798 CE, an index the Hub maintains for locating residents, which lists them by location on the Orbital and by nothing about their lives (Chiark Orbital Hub Records, 1798 CE). The survey asked these residents whether they regularly played any game with a ladder. Seventy-four said they did and were excluded, leaving 1,126 eligible, and 670 of these replied. We call them the comparison group. Everyone was told the survey was voluntary.

Standing was taken from the ladder books. For each player we used the place in the book for the game they played most, converted to a percentile among the residents assigned to that book, where 100 is the top place and 0 the bottom. We also used a top-decile indicator, defined before the surveys were sent as a percentile of 90 or more. The outcome was the number of invitations to gatherings not organised around a game or sport that the person reported receiving in the previous twelve months. Invitations are a natural unit of regard in a society where attendance is always a choice, and self-reported counts of this kind have been checked against hosts' own lists on Chiark (Maravile & Sosk, 1774 CE). Players also reported their weekly hours of play, the number of ladder games they played and their age band, in five bands suited to citizens who commonly live about three and a half to four centuries: under 60 years, 60–119, 120–199, 200–299 and 300 or more.

The outcome is a count whose variance in our data was 3.3 times its mean, so we fitted Poisson regression and judged it with cluster-robust standard errors, which stay valid under overdispersion if the model for the mean is right (Quillon, 1738 CE). For players, errors were clustered on the ladder book, 14 clusters in all, with a small-sample correction and confidence intervals from a t distribution on 13 degrees of freedom (4 for the five sport ladders alone and 8 for the nine table-game ladders). Model 1 contained standing alone. Model 2 added age band and ladder type, table or sport. Model 3 added weekly hours of play and games played. Model 4 replaced the continuous standing term of Model 3 with the top-decile indicator, and Model 5 entered both the continuous term and the indicator together. Standing enters in units of ten percentile points and effects are reported as incidence rate ratios (IRR). To compare players with the comparison group we fitted a Poisson model with age band and a player indicator, with heteroscedasticity-robust errors. All intervals are 95%.

3. Results

Of the 979 players invited, 842 responded (86.0%). The 14 books were assigned between 28 and 120 players each, with between 25 and 102 respondents per book. Of the respondents, 551 played table games and 291 (34.6%) played sports. They reported a mean of 6.88 hours of play a week (SD 3.41), and 411 (48.8%) played more than one ladder game. Hours of play and standing were weakly correlated (r = 0.20). The 842 players fell into age bands as 166, 192, 206, 183 and 95 from youngest to oldest, and the 670 comparison respondents as 139, 149, 146, 137 and 99. The two groups are therefore similar in age but not identical.

Players reported a mean of 6.66 invitations in the year (SD 4.70, median 6) and the comparison group a mean of 4.97 (SD 3.58, median 4). With age band held constant, players received 34% more invitations to gatherings unconnected with play (IRR 1.34, 1.25–1.44, p < .001).

Within the ladders, reported invitations rose with standing, though not steeply (Table 1). The upper half of the percentile range (428 players) reported a mean of 7.16 invitations and the lower half (414 players) 6.15. By quintile, means ran from 5.90 to 6.92 across the lower four and reached 7.82 in the top quintile, a descriptive pattern that the models below test.

In regression, each ten points of standing went with 3% more invitations in every model (Table 2). In Model 1, standing alone gave an IRR of 1.03 (1.02–1.05, p = .001). Adding age band and ladder type left it at 1.03 (1.02–1.05, p = .002), and adding weekly hours and games played gave 1.03 (1.01–1.05, p = .002). The unrounded coefficient of Model 3 implies a rate ratio of 1.35 (1.14–1.60) across the whole span of 100 points. Omitting any one ladder in turn left the ten-point IRR at 1.03 in every case. Weekly hours had an IRR of 1.07 per five hours (0.98–1.16, p = .14) and each additional ladder game 1.02 (0.96–1.09, p = .42), so neither was clearly associated with invitations once standing was in the model. Sport ladders had a higher rate than table-game ladders (IRR 1.38, 1.21–1.58), a difference that sits entirely at the level of the ladder, where we have five sport ladders against nine table-game ladders.

Model 4 looked at the top decile directly. Players in the top decile (92 players, 10.9%) reported more invitations than the other 750, with an IRR of 1.22 (1.10–1.36, p = .001). Averaged over the sample, the model predicts 7.94 invitations a year for a player placed in the top decile and 6.50 for one who is not, a difference of 1.44. Omitting any one ladder moved this IRR between 1.19 and 1.26. When both standing terms were fitted together (Model 5), the linear term was 1.03 per ten points (1.01–1.04, p = .007) and the top-decile term 1.08 (0.99–1.18, p = .08). The two terms are correlated (r = 0.54), and the data do not distinguish a steady rise in invitations from one that is steeper in the top decile. Within ladder type the ten-point IRR was 1.03 on table-game ladders (1.01–1.05, nine ladders, t on 8 degrees of freedom) and 1.03 on sport ladders (0.99–1.07, five ladders, t on 4 degrees of freedom), the latter with an interval that includes 1.

4. Discussion

Standing on a Chiark ladder does carry a little way beyond the game. Players who stood higher on their ladder reported more invitations to gatherings that had nothing to do with play, and the gradient survived adjustment for age, for the kind of game, for how much a person played and for how many games they played. It was small: a ten-point rise in standing went with about 3% more invitations, and the whole span from bottom to top with about 35%. Players in the top decile reported about 22% more invitations than the rest, which the model puts at about one and a half more in the year. Once a linear trend is allowed, however, the extra top-decile term is not distinguishable from 1, so the data do not tell a steady rise from one that is steeper in the top decile.

Two things stand out. The first is the size of the shift between players and the comparison group, 34%, which is about as large as the difference between the bottom and top of a ladder. People who play ladder games at all seem to be invited more widely than those who do not, and a person's place in the order adds roughly as much again across its full length. The second is how little weekly hours and games played explained once standing was in the model. If the gradient arose simply because frequent players are better known, we would expect hours of play to carry it. They did not do so clearly, though the estimate for hours is imprecise.

Ordinary social reasons could produce the whole pattern without any transfer of regard from game to wider life. Sociable people may play more, climb faster and receive more invitations for the same reason. People who are good at a game may be better company at it, and are invited to play, with some of the gatherings in our outcome shading into play. And the order of a ladder is public, so a neighbour who is choosing whom to ask may use it as a convenient clue to who is good to have around. Our data cannot tell these accounts apart.

What the data do support is the second of our two readings, in a modest form. Games are not wholly sealed from the rest of life: a public order of skill goes with a little more regard outside play, even in a society with nothing to enforce it. We have no comparable figure from the civilisations Contact describes, so we say nothing about how much tighter the link is there.

5. Limitations

Invitations are self-reported counts over twelve months, and recall will be imperfect. The validation we rely on was done on Chiark but with a different sample and a different decade (Maravile & Sosk, 1774 CE). The invitation window ran back from the survey date, while standing was fixed at the close of the 1797 CE season, so the two overlap in time and we cannot show that standing came first. Regard from a ladder could as easily follow from being invited widely as lead to it.

The study has 14 clusters. Cluster-robust errors with so few clusters can be too narrow even with the small-sample correction we applied, and the contrast between sport and table ladders is carried by five and nine ladders respectively and cannot be separated from other differences between clubs. Fourteen percent of invited players and 40.5% of eligible comparison residents did not reply, and we have no information on how they differ from those who did. The 52 entries we could not resolve were absent from the analysis, and their standing is unknown. We computed percentiles among resolved residents only.

Finally, the sample is one Orbital in one decade. Practice on other Orbitals, aboard General Systems Vehicles and for other games may differ, and the figures reported here are estimates for the people we reached, not for the Culture as a whole. We report the continuous and top-decile terms, singly and together, because the first was fixed as our main model and the second as a secondary one; neither was chosen after seeing the other.