This article describes the subscription model that Sociological Research Online has adopted, and reports how it is implemented using current technology. The main aim of the model was to provide a service that required little administration or maintenance. The model adopted fits the brief well.
Sociological Research Online's Subscription Model
Implementing the Model
During 1998 and 1999, the Management Board of Sociological Research Online [ 1] considered ways of collecting subscriptions from readers. The journal's funding ceased near the end of 1998; it was important that an income stream would be identified to secure the future of the journal. Some alternatives to the traditional subscription system were sought but subscriptions were eventually felt to be the most realistic way forward. One of the attractions of a subscription system is that payment arrives 'up front', before the issues have been published.
By far the largest cost of Sociological Research Online is staffing [ 2]. The journal employs a Publishing Assistant to work two days a week. The Assistant is responsible for day-to-day administration (which includes taking articles through the review process) and for publishing the journal on a quarterly basis. The Assistant also answers inquiries, such as reproduction requests, and maintains the journal's Web site and various online services.
Many electronic journals are run on a not-for-profit model whereby a single individual or small team work to produce an independent journal. While this works well for small journals in tightly focussed sub-disciplines and for low volume inter-disciplinary titles, running a larger operation that aims to compete against leading print journals requires a more secure operating environment. The work would be too onerous for a small team of academics already having too many other demands on their time. In addition, a 'publisher' adds value to a journal by offering marketing expertise and adding 'brand' support. Publishers are also better placed to offer a guarantee of long term security than individual academics.
Sociological Research Online's Subscription Model
Although based on a traditional subscription model, Sociological Research Online has introduced some novel features. First, it was important not to add too heavy or complex an administrative burden so that the Publishing Assistant could handle subscriptions along with all other aspects of the job. Second, it should be possible to charge large institutions for more than one subscription in the same way that they have to purchase several subscriptions of a print journal in order to cover all of their libraries. Third, a provision should be made for 'lone scholars' having no institutional affiliation, institutions in developing countries, and academics at institutions which simply will not subscribe to electronic-only materials or this journal in particular.
The solution that Sociological Research Online adopted meets these criteria. The Management Board decided that the journal would be offered to institutions for an annual subscription fee, but would continue to be freely available to individuals accessing it from dial-up Internet accounts. Free subscriptions would also be given to countries whose their GDP fell below a certain threshold. A subscription would permit access for a specific network. Institutions spread across many campuses or ones so large that they require multiple networks therefore would have to purchase multiple subscriptions.
In addition to giving 'lone scholars' continued access to the journal, allowing free access to dial-up Internet account holders avoids the troubles of handling personal subscriptions. In turn, the Publishing Assistant need only deal with a handful of subscription agents. This alleviates the need to use username/password authentication - widely acknowledged as being onerous to manage and placing an unfair burden on subscribing libraries who have to offer first line of support for problems, and who must manage the distribution of login details.
Print journals generally start with a high number of personal subscribers which eventually translate into institutional ones over the first five or so years. Not having personal subscribers puts Sociological Research Online at a slight disadvantage, but this is more than compensated by the fact that the journal is still free to individuals, and that the first three volumes of the journal remain free to everyone. It is hoped that this free access will lead to recommendations for subscriptions to institutional libraries.
Implementing the Model
The subscription model is reasonably straightforward to implement and enforce since it takes advantage of existing technology. There are three key parts: user authentication; denial of access to non-subscribers; and, a request mechanism for free access from dial-up networks.
The procedure for network authentication is termed Internet Protocol (IP) authentication. Every computer on the Internet is assigned an IP address, which may be fixed (static IP) or assigned when logging on to a network (dynamic IP). Regardless, every client computer has an IP number that identifies that client uniquely from any other one. Commercial publishers are often wary of IP authentication, citing the ease with which computers can 'spoof' (ie. conceal) which network they are on. In practice, such spoofing is far more difficult than is usually credited and any likely disadvantage of this system is more than overcome by the ease with which authenticated users can access the journal.
IP addresses are usually written as four sets of numbers (known as dotted decimal, or dotted quad notation), each having a value between 0 and 255. The four numbers represent a single 32 bit number (32 binary 1s and 0s). Thus the IP address 126.96.36.199 distinguishes one computer from all others on the Internet. Moreover, the first one, two, or three sets of numbers define a network that the computer is on, and the maximum size (and hence Class) of the network. If the first number is in the range 0 - 127, the network is known as a Class A network. If the first number falls in the range 128 - 191 the network is a Class B network, and if it is within 192 - 221, the network is of Class C [ 3].
On a Class A network, only the first number is significant in identifying the network and the remaining three numbers can be assigned by the owner of the network. Such networks can include very many clients but they are rare (there are only 128 in total). Class A networks are reserved for the U.S. Government and extremely large Internet providers. The first two numbers are significant on a Class B network, with the remaining two left to be assigned by the network owner. Class B networks are commonly adopted by large institutions such as commercial organisations and universities. Class C networks are denoted by their first three numbers and only the final one can be assigned by the network owner. There are many more Class C networks than Class B, and each has a maximum of 254 (1 - 254) clients. Smaller universities, organisations, and Internet Service Providers (ISPs may be allocated several Class C networks).
One subscription to Sociological Research Online, which costs £100 (about US$150, plus VAT where applicable) will allow access to a single Class B or equivalent network. Very large universities may have several Class B networks and will be required to purchase a subscription for each one from which access to Sociological Research Online is required. Two subscriptions would be needed for one Class B and a number of Class C networks. Because of the way that Class A, B, and C networks are allocated, it is possible to see at once which Class a network falls into, and it is easy to determine how many subscriptions are required for access to all of an institution's specified networks.
Subscription agents are poor at passing on network information with the original subscription order. This means that the Publishing Assistant often has to determine which networks are permitted access. This is reasonably straightforward using 'nslookup' and 'whois' services designed to query Internet networks. If you do not have local access to these, then various sites allow online look-ups [ 4]. First, one network must be identified. This information can then be used to determine any other networks belonging to the same institution, and then the subscription is set up. The task can be broken down into steps:
- Identify a Web page belonging to the institution. University home pages can usually be reached at http://www.<UNIVERSITY>.ac.uk for the U.K. American universities usually have the prefix '.edu' and the same abbreviation is used for many other countries, followed by the country code, eg '.edu.nz', '.edu.za'. European universities usually omit '.edu' which makes them harder to trace. If guesswork fails, university Web sites can be located rapidly using ordinary Web search engines.
- Take the Web address (URL) from step 1 and use it to form a query using the tool 'nslookup':
> nslookup www.cam.ac.uk
- Use the second Unix tool, 'whois' to see what other networks are allocated to the same or affiliated institutions:
> whois 188.8.131.52
University of Cambridge (NET-CAM-AC-UK)
Computing Service, New Museums Site, Pembroke Street
Cambridge, CB2 3QG GB
Domain System inverse mapping provided by:
Record last updated on 07-Jan-1999.
Database last updated on 20-Jun-2000 17:58:26 EDT.
- Allow access to users from these networks.
The resulting record confirms that networks 131.111, 128.232 and 129.169 are assigned to the University of Cambridge. These networks may be allowed (or denied) access to the journal.
Restricting Access to Web Pages
Access restriction to subscribers is handled by the Web server. Sociological Research Online is served by Apache, one of the most widely used Web servers. The server can be configured to follow directives placed in files named (by default) '.htaccess'. These files are located within directories of the Web structure and affect that and any lower directories. The directives can specify which hosts to allow or deny access to, which files are secured and which are open, and whether usernames and passwords are required [ 5].
To add a subscription for a new institution, the IP network details are added into the '.htaccess' file for each relevant directory. This file must also contain directives specifying which files must be universally accessible. Such files may include content listings, images, and other informational files.
Sociological Research Online uses an automated system developed by EPRESS [ 6] whereby subscription details are stored in an online database. Each time a subscription record is added or modified, the update procedure recreates the '.htaccess' files in all the directories. Details of free access files are stored in the database in a separate table, as are any other special rules to be added into the access restriction. Such a live system has benefits when trying to troubleshoot access problems from a subscribed institution. Changes can be made with immediate effect and the user can check whether problems have been solved in conversation with the client over the telephone.
If a non-subscriber attempts to access the full text of articles, the Web server is configured to deliver a customised error message. For Sociological Research Online, a script returns the abstract and keywords of the article. Beneath this is text explaining that they cannot access the whole article because their network is not subscribed. They are encouraged to recommend their library to subscribe if they are reading from an institutional network otherwise they are asked to submit a request for free access.
Readers browsing the journal from a network that has already been subscribed will receive the whole article without even realising that a subscription mechanism is in place. Such seamless access is an important feature of the journal.
Access to ISP Account Holders
The term 'dial-up account holders' has so far been used for those accessing the Internet and the journal through a third-party Internet Service Provider. Indeed these readers may in fact be from small organisations who have Internet access provided by a generic ISP but as it is impossible to make a distinction, it is not possible to bar them.
ISP users can submit details via a form that sends a request for access to the journal. They are asked to type in their name, e-mail address, and the name of their ISP. When they click 'submit', their details are placed directly in the subscription database. In addition to the three fields requested, the receiving script also logs the IP address of the computer they submitted from, and, if registered, the host name of the client. This name is one that maps onto the IP number and may give further clues about the originating network.
Every request for free access from a dial-up account is checked manually to ensure that the request has come from a proper Internet Service Provider and not an institutional account. 'Whois' can be used in conjunction with a visit to the provider's home page to verify the request. As the request form has already logged the IP number of the requesting client, the network can quickly be queried using 'Whois' (step 3 of the scheme outlined above). It is therefore easy to establish whether this network belongs to an ISP or the request is from an institutional network. If the request is valid, the subscription is moved to 'active' by the Publishing Assistant, and any additional networks owned by the ISP (also returned by 'Whois') can be added to the record.
The model adopted by Sociological Research Online is as perfect a solution as could be envisaged. One criticism levelled by a librarian was that there is no provision for an academic from a subscribing institution on sabbatical leave at a non-subscribing institution to gain access to the journal. Although print journals suffer this problem, Sociological Research Online does not. It should be possible for a wandering academic to log into their own institutional network from anywhere in the world using Telnet and then access the journal with a browser such as Lynx. Any limitation is in the capability of the individual reader rather than the technology.
The model satisfies distant learners and homeworkers who can either dial-up through their university network and, if subscribed, will be granted access or can register their ISP and gain free access through that route.
Most of the problems associated with the practical side of the subscription system arise from subscription agents. Most libraries place all their journal and book orders through one or two agents who ensure that all orders are fulfilled, and subscriptions are properly established. Not all agents are yet experienced in handling electronic-only material. For example, one agent cancelled a client's subscription because they did not realise that the journal was electronic (they requested a paper subscription). The agent should have been able to spot this error while they were processing the order.
One problem at the journal end is that it shares a single bank account with the whole of the University of Surrey which hosts it. One agent frequently pays cheques for new subscriptions directly into the bank account; this makes it difficult to match a specific payment to an order, particularly as the paperwork often arrives after the cheque has been delivered.
There is a technical problem associated with the use of caches. U.K. institutions are particularly affected as they are encouraged to use the U.K. national cache provided for the academic community. However, the cache is set up in such a way that requests originating from a university, but directed through the cache, do not pass on the details of the original client. This means that if clients from subscribed institutions use the cache, their request appears to come from the cache and, because it cannot be traced back to the original institution, access is denied. In order to prevent this, each institution has to set its local cache to bypass the national cache for requests from Sociological Research Online. Unfortunately, it is not possible to make a single change at the national level to prevent this problem. A partial solution for users may be to turn their browser's cache off. Whether this will be successful depends on how the institution is set up to handle Internet requests. All calls may automatically be intercepted and directed to a local cache which in turn may point to an external one, in which case this fix will not work. In the longer term, users must ask their network support staff to prevent requests for Sociological Research Online from being passed to the cache.
Sociological Research Online is of the opinion that subscriptions are a necessary evil. In order to maintain the quality and secure the future of the journal, some income is required. However, the journal does not have to seek profits like those journals published commercially. This situation allows Sociological Research Online certain freedom to remain freely available to selected readers (i.e. those accessing via ISP). Over time such access may work in favour of the journal if it encourages widespread readership which in turn may translate into a greater number of institutional subscriptions.
The main concern of the subscription system, however, was to introduce a system that required as little management as possible. In almost every respect, the model developed by Sociological Research Online fits the brief well.
About the Author
Stuart Peters is a Research Fellow at the University of Surrey, UK. Between 1995 and 1998 he was publishing Sociological Research Online, funded by the Electronic Libraries Programme (eLib). In 1998 he started EPRESS, a two-year project also funded by eLib, which has developed a software infrastructure for publishing electronic academic journals.
1. Sociological Research Online: http://www.socresonline.org.uk/
2. For more information on Sociological Research Online, see Peters & Gilbert (1997).
3. Class C IP networks above 221 are reserved for a variety of uses.
4. For example, see http://dns411.com/
5. Consult the Apache documentation ( http://www.apache.org) for a more detailed account of access configuration.
6. Electronic Publishing Resource Service: http://www.epress.ac.uk/ See also Peters (1999; 2000).
S.M. Peters and G.N. Gilbert, 1997. "The Electronic Alternative: Sociological Research Online," Learned Publishing, volume 10, number 4, pp. 339-343.
S.M. Peters, 1999. "EPRESS: Changing the way electronic journals work," VINE, volume 111, pp. 31-36.
S.M. Peters, 2000. "EPRESS: Scaling Up Electronic Journal Production," Ariadne, issue 23, at http://www.ariadne.ac.uk/issue23/epress/
Paper received 26 June 2000; accepted 29 August 2000.
Copyright ©2000, First Monday
Presenting a Successful Electronic Journal Subscription Model by Stuart Peters
First Monday, volume 5, number 9 (September 2000),