This is the first post in a multi-part series about viewdata. A few members of the hackspace are working on a project to connect a ZX Spectrum with Prism VTX 5000 modem to an emulation of a viewdata system, providing a live demonstration experience.
Back in the early ’80s, before the world wide web was a thing, there was something a little like Teletext called Viewdata.
Viewdata pages look a lot like Teletext – they share the same frame formats – and viewdata terminals likely used the same IC to produce the video signal (The Mullard SAA5050, also used in the BBC Micro for its text-only mode). While teletext pages are transmitted in the blanking interval of a TV signal, in rotation so a receiver needs to wait for the right page, viewdata operates over a telephone line and so is more interactive.
When a subscriber wanted to log onto the service, they dialed the provider they wanted to access – for example Prestel – and provided their ten-digit ID number and password to log in.
Once logged on, the user was presented with the main page, and navigated using the number keys and enter. Pages were generally arranged hierarchically, so page 1234 would be below 123, and is reached from that page by pressing ‘4’. Like Teletext, viewdata pages could consist of multiple frames, so pages longer than one frame could be constructed.
Companies who wanted to host information on a viewdata service would pay the information provider (IP); this would involve a payment for a particular length of prefix (shorter prefixes would typically be more expensive), and also per page hosted. They could edit frames using special editing software, and then uploaded these to the IP. Frames weren’t restricted to being completely passive – “submission” frames could be authored, allowing people to order goods or fill in forms. Frames could also be marked with a cost, allowing companies to collect money against a user’s account when they accessed the information.
Often, information providers would have a mailbox system, allowing users to send each other messages, which could be retrieved and read when the user logged in. Some terminals would allow these, as well as other frames, to be retrieved and stored locally, so they could be read offline without the user keeping the line open.
As well as centralised information providers, such as BT’s Prestel, software later emerged allowing people to host their own viewdata systems – often bulletin boards – or set up peer-to-peer connections. Peer-to-peer connections allowed users to phone each other, and as long as they set their modems correctly, information could be transferred half-duplex. Bulletin boards could be run on systems such as the BBC Micro, and would present themselves in the same manner as an IP.
The modem standard – v.23 – used for viewdata is asymmetric. The forward channel (download) runs at 1200 bits per second (120 characters per second taking framing into account), while the reverse channel (upload) runs at 75 bits per second (7.5 characters per second). A terminal – apart from any storage functionality – is relatively “dumb”, and transmits key presses immediately. The information provider’s system reacts as needed – whether this means sending a new frame, or echoing characters typed into a response frame. The slow reverse channel is still generally fast enough to transmit the user’s keypresses, and is fairly noise-resistant because of the low bandwidth.
There are sites which contain demonstrations of some viewdata systems – https://www.viewdata.org.uk/ is a good example of this. Follow the “Logon Now” link to access an example.