Backgammon on the Net Survey

Introduction and Objectives

There has been much discussion recently (July 1996) in regarding the architecture of the recently-introduced Internet backgammon servers compared with that of FIBS (First Internet Backgammon Server). This survey is an attempt to discover backgammon players' server preferences and intentions, and to understand how their own local platform architecture will affect their use of these servers. It also attempts to analyse willingness of backgammon players to pay for access to these servers.

This survey was based on a sample of 60 respondents to a questionnaire posted to in July 1996. All respondents were self-selecting. The major characteristics of respondents were that they play backgammon on the Internet and they read the newsgroup

Four respondents had a special interest in one or other of the servers; two of them (by my understanding) had a vested financial interest.

This survey was conducted and authored by Marina Smith ( with assistance from me (mail form). Background section and HTML markup by me.

The survey is also available by email from (State whether you require a Microsoft Word 2.0 or an ASCII text file).

22nd July 1996


A central issue of the debate surrounding the new backgammon servers is openness of architecture. The terms 'openness' and 'architecture' may mean little, or seem of little importance to those outside the computer industry, but they are fundamental concepts that affect one's ability to use any Internet resource. Anyone expecting to use Internet resources in the future may benefit, therefore, from an understanding of these concepts.

There are three simple measures of 'openness' of any Internet server, including backgammon servers.

Table 1: Openness of Server
Description Protocol type used Availability of protocol spec.
Open Existing Internet protocol Public domain
Semi-open Proprietary Public domain
Closed Proprietary Unpublished

Internet backgammon servers currently fall into two categories: open and closed (there are currently no semi-open backgammon servers to my knowledge.) Note that I am talking about transport protocols, not data formats.

Open Servers

An open server makes use only of an existing Internet protocol. It can therefore be accessed by any client that satisfies two requirements:
  1. It can run any third-party client software designed for that protocol
  2. It is capable of supporting that level of Internet connectivity
The two requirements are not necessarily mutual. For example, both the Usenet protocols (nntp and uucp) and World Wide Web protocol (http) are public domain and clients are available for most platforms. But whereas the character- based Usenet can be access by any Internet access device (e.g. PDA), the multimedia Web can typically be used satisfactorily only by devices powerful enough to support multimedia (e.g. desktop computer).

In the case of open backgammon servers, the protocol is telnet. As telnet is a simple, character-based interface that requires negligible local processing resources, telnet clients are available for most if not all Internet access devices. Therefore, most if not all Internet access devices can be used to connect to open backgammon servers.

The convenience of being able to pick up any Internet device and connect to any open server is countered by the unsophisticated nature of a generic interface and the requirement, often seen these days as regressive, to use a command line interface (these apply to telnet). Neither restricts one's play or withholds functionality. However, as the protocol is in the public domain, any interested developer is able to write a specialised client for any platform, either in response to requests or for other reasons, for a prettier interface at the expense of portability.

As the protocol is standard and public domain, other servers can use the same protocol. This means the same clients (either generic [telnet] or platform-specific [graphical backgammon interface]) can be used on different servers. For example, NOBS (New Other Backgammon Server) was written by someone unconnected to FIBS to mimic FIBS, for the times when FIBS is unavailable, and can be used with any FIBS client.

As a sidenote, JavaGammon, a backgammon server that uses the Web's http protocol and supplies its own client as a Java applet is currently under development. This satisfies the principle requirements of openness, but currently cannot be said to be open to all, for the simple reason that Java itself is an immature technology that requires a relatively high minimum level of local processing resources. In time, Java is expected by many to become the default method of deploying and executing applications, and it will be supported by all types of client.

An example of an open server is FIBS. It runs over a plain telnet session, and several third-party developers have written platform-specific graphical clients.



Closed Servers

A closed server makes use of proprietary protocols developed by the designer of that server. As the protocol is a single-function protocol designed only for the purpose of connecting to a specific server, no existing client software can be used to connect to that server (unless special provision is made to accept, for example, incoming telnet connections, which is currently not the case). A specialised client must be developed for every hardware and operating system combination before that client can connect to the server. As the protocol used by each closed server is currently incompatible with those of all other servers, a client platform must have access to a different client software package for each closed server to which it is to connect.

The number of client platforms, as we currently understand the term, is considerable, comprising, among others, desktop computers, portables, home computers, and dumb terminals. However, when we define the term 'client platform' to mean any device which can be used as a client to an Internet server or application, the number increases many times. Such devices as game consoles, set-top boxes, Network Computers, and personal organisers are included in this definition and, for the purposes of accessing Internet resources, are of equal status as traditional computer systems. Some believe that such devices will open Internet access to 'the masses' and will outnumber traditional platforms for Internet access within five years.

To write, test and support a backgammon server's client interface for all of these platforms is beyond any single developer or organisation. (Don't forget that within all the above categories will be many different products based on different architectures, each of which needs special attention and support.) Therefore, one of two scenarios will emerge:

  1. Not all client platforms will be supported
  2. Developers concerned with a subset of specific platforms will develop clients for those platforms
This is where the difference between semi-open and closed servers arises. A semi-open server's protocol is in the public domain. Although a user cannot access the server without a specialised client (unlike open servers), the development of specialised clients is open to all. In the same way as individuals and organisations develop, for example, Web clients for many different platforms, even obscure ones, so clients for many different platforms would be developed for connecting to a semi-open server.

A closed server's protocol, however, is unpublished. Clients for different platforms can be developed only by the owner of the server or by a licensee of the protocol. If the protocol is not licensed, it is inevitable that most types of platform will not be supported by a client - the server's owner cannot possibly write clients for all platforms. If the protocol is licensed, then potentially more types of platform will be targeted by client developers, but licensees will restrict themselves to high-volume platforms in order to recoup their licensing fees, which we must assume would exist in the case of a closed server. (If no license fee is charged, one could question why the protocol specification is not in the public domain.)

Closed servers tend to feature a strong level of integration between client and server. Functionality provided by the server typically has to be catered for by a mirrored 'function request' feature in the client. Therefore, as the server increases in functionality, so the client has to be updated to be able to access that functionality. Some new server functions can potentially be accessed by generic elements of the client interface (for example, simple context-sensitive 'yes' and 'no' buttons), but many can not. For this reason, users will be required to upgrade their client to take advantage of new features. This incurs either a complete download and reinstallation of the client, or a download and application of a software patch. With current graphical operating systems, the size of file to be downloaded can be considerable.

Examples of closed servers include NetGammon, Games Grid and Outland. Each uses a proprietary and as yet unpublished protocol and requires a client to be installed. Each is currently restricted to either a single client platform or a very small set of platforms.



Survey Results

Hardware Access Equipment

There is much discussion about minimum requirements needed to play on some of the backgammon servers, the point having been made that for some of them, there is no minimum. The reason for eliciting the hardware by which people access the Internet is that some backgammon servers require a minimum specification of client platform. Computers used by respondents to access the Internet (not just to play backgammon) are as follows.

Table 2: Computers Used
Platform Percentage
PC 58
UNIX workstation 30
Apple Macintosh 11
vt100 terminal 1
Sample: 73

Thirteen people connect to the Internet using two different platforms, which accounts for this sample being larger than the overall survey sample of 60. Similar differences in sample size occur throughout the rest of the survey.

Table 3: Access Devices
Device Percentage
Modem 70
Leased line 24
Sample: 74

The percentages do not add up to 100 due to rounding. This effect occurs elsewhere in the survey.

It should be noted that a leased line is one that keeps a permanent connection open whether there is traffic or not. An ISDN line is dropped when not in use in most countries. In the U.S.A. this is not necessarily the case - since local calls are often free of charge or there is a small charge for the initial connection only, it is more convenient and/or more economical to stay connected, in which case an ISDN line acts like a leased line. Also in the U.S.A., some so-called ISDN lines are actually 'switched 56' lines running at 56K bps; I have not differentiated here as it makes little difference for Internet access. I have also not made any differentiation between 64K bps access and higher speeds (some people have much higher speeds - up to 1.544M bps) since figures are often unknown with leased lines and because for games-playing purposes, 64K bps is enough to counter any access lag, most lag occurring at intermediate bottlenecks in the Internet. In fact, 1,200 baud, the lowest speed found in this survey, proves just about adequate for playing backgammon.

Table 4: Modem Speeds
Speed Percentage
1,200 baud 2
2,400 baud 2
9.6K bps 2
14.4/19.2K bps 44
28.8/33.6K bps 50
Sample: 48

Table 5: Place of Play
Location Percentage
Home 90
Work/university 55
Both home and work/university 48
Sample: 60

Online Connection Charges

The questions as to payment for online time were intended to help see a clear picture of who already pays for every minute they are connected to the Internet and who therefore may not be willing to pay more to play backgammon on the net. I have assumed that everybody playing from home pays a flat rate charge both to a telco and to an ISP (Internet service provider). Since these connection charges would be paid anyway for matters unrelated to backgammon, I have discounted these charges. Unfortunately, these questions were misunderstood by some respondents, who gave positive answers for flat rate charges. I can therefore only give results for those who definitely do not pay, but it is clear that these figures may be higher.

I found that, given this caveat, only 21% of people who play from home pay their telco for online time and only 23% of people who play from home pay their ISP for online time. At least 61% of people who play from home pay neither their telco nor their ISP, i.e. their online time is unmetered.

It should be noted that residents of many U.S. states enjoy unmetered local calls. While some are charged instead by their ISP for online time, it is clear that many are not. These figures may come as a surprise to some outside the U.S.A. As most of the survey respondents were from the U.S.A. (although I cannot quantify the amount, due to the non-geographical nature of many domain names), I would expect these figures to change in the future as the U.S.A. accounts for a decreasing proportion of the worldwide Internet population.

Backgammon Servers

There are currently six backgammon servers on the Internet.

Table 6: Servers
Server Clients Supported Protocol * Payment
FIBS Any client Open None **
NOBS Any client Open None
NetGammon PC, Windows 3.1 upwards Closed Expected
Games Grid PC, Windows 3.11 upwards Closed Expected
Outland Apple Macintosh Closed Expected
JavaGammon Any client Open None
* Using the definitions given in the Background section
** Voluntary donations are accepted by a third party unconnected to FIBS to pay for current FIBS running costs

All but two respondents currently play backgammon interactively on one or other (often more than one) of the servers. Two do not have access at present. One person plays by email, but also plays on a server.

Table 7: Backgammon Server Preference
Server Percentage
Games Grid 10
NetGammon 2
Outland 0
Sample: 60

NOBS was not given as a preferred server, as it is used primarily as a backup server for when FIBS is unavailable. While NOBS is very similar to FIBS in functionality, it is not always accessible, and is not intended to compete with FIBS.

Respondents were asked the reasons for their server preferences. Only FIBS users supplied enough comments to measure meaningfully. Games Grid users supplied 11 comments which, although not defensible statistically, also appear below to give a rough idea of their preferences.

Table 8: Reasons for Preferring FIBS
Reason Percentage
Open architecture 39
Strong players 16
Atmosphere 16
Unawareness of other servers 8
Free 8
Functionality 8
Interface 4
Ease of connection 2
Sample: 51 identifiable mentions

Table 9: Reasons for Preferring Games Grid
Reason Percentage
Speed 36
Interface 27
Functionality 18
Atmosphere 9
Ease of connection 9
Sample: 11 identifiable mentions

Sample quotes:

  • "[Games Grid] Faster server response. Nice people. No server crashes"
  • "Love FIBS, can't use NOBS, can't find NetGammon, and the idiots from GG [Games Grid] have made sure that I will never want to use their 'service'. GG have given themselves so much negative advertising that no amount of marketing would make me use GG"
  • "[FIBS] A good rating system, world-class players, and an open architecture"
  • "In order: Games Grid, FIBS, NetGammon. If limited to one choice: Games Grid for its speed and user-friendliness"
  • "[FIBS] At home: the other clients are too heavy for my PC. At work: FIBS can be used just with a plain 'telnet'"
  • "FIBS, glorious FIBS. It's free, flexible, friendly and fun"
  • "[FIBS] More players, more 'life', better talking commands, and easy to get an overview of available players when you're playing yourself"
Table 10: Interface Type
Interface Percentage
GUI 75
telnet 25
Sample: 65

All the current servers except FIBS and NOBS require proprietary GUIs (graphical user interfaces) to be installed on the client platform. A quarter of current Internet backgammon players are therefore unable to play on other servers using their current client software.


Payment for access is not currently required by any backgammon server, but is likely to be demanded in the near future (see Table 6 for the servers which are expected to carry a subscription fee). The longest running server, FIBS, was forced to change location in June 1996 and many players, concerned that it would disappear, offered money to help fund a suitable site. FIBS was rehoused at an ISP's site and is currently paid for by two players, with voluntary donations from others. FIBS itself is still, and will remain, free to users.

Table 11: All Respondents
Willingness to Pay Percentage
Yes 60
No 27
Undecided 13
Sample: 60

Table 12: Unmetered Internet Users
Willingness to Pay Percentage
Yes 56
No 22
Undecided 12
Sample: 34

Table 13: Metered Internet Users
Willingness to Pay Percentage
Yes 61
No 22
Undecided 17
Sample: 23

It may be surprising that unmetered Internet users (those who pay neither telco nor ISP for online time) are actually less willing to pay to play on a backgammon server than are metered Internet users. Perhaps this might be explained by the possibility that a number of people who access the Internet for free do so only because it is free. This body of people may be less likely to pay to access a backgammon server than those who currently pay for Internet access. Perhaps, realistically, it might also be explained by the small sample sizes involved.

Table 14: By Preferred Server
Willingness to Pay FIBS Games Grid
Yes (%) 56 100
No (%) 29 0
Undecided (%) 15 0
Sample size 52 6

As only one person gave NetGammon as preferred server, NetGammon was excluded from this table. It is highly questionable, too, whether the six people who gave Games Grid as their preferred server make up a statistically valid sample. Their responses are given here not as a statistical comparison with those of FIBS users, but as an indication of their apparent willingness to pay to play.

Table 15: Reasons for Unwillingness to Pay
Reason Percentage
Continued availability of free server(s) 42
Goes against Internet culture/philosophy 21
Concerns about stability and availability 13
Inadequate functionality 8
Would rather pay for real-life backgammon 8
Don't play enough to warrant a fee 4
Too much trouble 4
Sample: 24 identifiable mentions

Sample quotes:

  • "A free one would open up as soon as a pay service started"
  • "I'm happy with FIBS, and it's free. Why throw money away?"
  • "Would prefer not to pay - servers seem to be unstable and I would not want to pay for lag time"
  • "I can already 'pay to play' in chouettes and 'pay to learn' with Jellyfish"
  • "Able to pay, but somewhat unwilling. Seems to invoke an unfriendly/greedy/regimented spirit into an otherwise friendly/sharing/free environment"
  • "While willing to support FIBS, I would be unwilling to have to pay for a service which FIBS supplies (will supply) for free. And a service which would require me to use certain specific software, etc."


'Spirit of the Net' vs Commercialism

The differences of opinion as to choice of backgammon server centres partly around a question of community spirit (mainly from FIBS adherents) and functionality (most Games Grid and some FIBS users). This can be seen in the typical replies to the qualitative questions on the topics preferred server and willingness to pay. Responses here mentioned, on the community spirit side, such factors as atmosphere, nice people, openness of FIBS to any client software. Some of the answers to do with functionality assumed that many of the characteristics of traditional, standalone Windows applications would be, and should be, replicated on the backgammon servers. This split reflects the changing nature of the Internet. It is growing from being a clique of technically-minded people who have become imbued with the values of the Internet as it has always been - open, shared, free, anarchic - into a wider world of less technically knowledgeable people. These are often people who hold, instead of these 'Internet' values, a purely commercial sense of what is appropriate in the applications they use on the Internet.

Free vs Commercial Servers

The ideas that lie behind the two main servers (FIBS and Games Grid) are shown by the responses to this survey. FIBS was designed as a student's project and has been on the Internet since June 1992. It embodies the spirit of the Internet in that it is free, open, and the product of a labour of love rather than a commercial venture. Given its age, a far greater number of respondents can therefore be expected to write about FIBS. This is reflected in the statistics given in the survey. Games Grid, a recent addition to the Internet, has been designed to give "good value for money" (one may read: it has been designed to make money) in that it has a fully designed interface and is fully functional for commercial use. It does not yet charge a subscription, but it is clear that eventually it expects to do so. The respondents to the survey appreciate this approach; if its adherents are indeed comprised of the wider type of Internet user, then this is only to be expected. This is a difficult assumption to make, but the survey here shows pointers that this may be true, given the comments made in showing preferences.

Open vs Closed Architecture

The designers of FIBS and Games Grid have made very different assumptions about the people who will access their servers. FIBS allows anyone, with any hardware or software, to join. Games Grid has looked for the most popular platform (PC with Windows 3.11 or higher) now in use in people's homes, and initially supports only that. I find that indeed the majority of Internet backgammon players play from home with a PC and modem, which gives Games Grid the possibility of at least short-term commercial stability, but excludes most potential users. The operators of FIBS have stated that there will be no charge for use of their server; other servers (with the assumed exceptions of NOBS and JavaGammon) will start charging one day. Users of such servers will therefore have to pay their telcos for access to the Internet (flat rate plus online charge in many cases), a fee to the ISPs (mainly flat rate), plus a subscription charge to the backgammon server of choice. For many players, this is intolerable, especially since there are free servers in existence. In fact, I found that around two thirds of players incur no connection time charges from their telco or ISP, yet this subset is slightly less willing to pay to play. Concerns were expressed not only about issues of free servers and Internet community spirit, but also about practical considerations of stability and availability, reflecting the immaturity of backgammon on the Internet as a commercial consideration.


Survey Questionnaire

  1. What hardware platform do you use to access the Internet? (e.g. 486 PC, Mac, UNIX workstation)
  2. What communications type do you use? (e.g. modem, leased line)
  3. What speed of access do you have? (If you do not know, just say so)
  4. Do you play from home or from work/university?
  5. Do you pay the phone company for your online time?
  6. Do you pay your Internet service provider for online time?
  7. Do you play backgammon on the net?
  8. If so, which server do you prefer? (FIBS, NOBS, NetGammon, Games Grid - if you play by email, just answer "email")
  9. Why? (Answer in 12 words or less, please)
  10. Would you be willing/able to pay to play?
  11. If able but unwilling, why not? (Answer in 12 words or less please)
  12. Do you use a graphical interface?

Back to the Netadelica home page