The ALL-IN-1 Mail Filter: Forerunner of modern email rules processing

Email systems had a different ebb and flow twenty-five years ago. An average mailbox received perhaps 10-15 messages daily and the messages were simpler with fewer attachments and properties. Most messages were in the order of one printed page or less, or roughly 2KB of ASCII text. Some email systems, such as DEC’s ALL-IN-1 (an early timeshared Office Automation system that had email capabilities), allowed users to compose message text with more complex editors (WPS-PLUS or WordPerfect) but the lingua franca was plain text as that’s all that you could reliably guarantee a recipient to be able to read if the message went “off-server”. When printed or displayed on-screen, a message usually still boasted that it was an “Interoffice Memorandum” or similar official-style communication. In the early days of email, we were still in the process of replacing typed memos and in many companies it was important for email to mimic the style and appearance of typed output.

Today’s users probably process ten times the mail traffic that we handled then. But even so, it seemed like a good idea to come up with methods to help users to process email automatically by allowing them to create a set of rules that prioritized messages by reference to keywords and other message properties and then take an action on messages to refile them into a folder or leave them in the Inbox for the user to process. This project became the ALL-IN-1 Mail Filter, an idea that’s described in U.S. patent 5377354 “Method and system for sorting and prioritizing electronic mail messages” granted on December 27, 1994. I am a co-inventor of this patent.

In patent-speak (written so only a lawyer could love the text), the patent covers:

A method and apparatus for prioritizing a plurality of incoming electronic mail messages for a user uses a user created and modified rules-control (12) which is stored in a rules-store (12). Incoming messages are stored in a message store (11) and are screened individually by a rules test unit (13). The rules-test unit has a comparator (52) which matches keywords which are chosen by the user while creating the rules, add supplies signals to an action list unit (54). By applying the user created rules for deciding which messages constitute the priority messages for the user, a priority assigning unit (45) within an action portion (35B) of the rules-store (12) assigns a priority number (say from 1 to 5, 1 being the highest priority for example) to each screened message. Responsive to the assigned priority number of the screened message, the message is sent to a main folder store or forwarded or put away as appropriate. The user created rules can be modified by the user using a conventional keyboard.

In human-speak, what we did was to create an ALL-IN-1 option to allow users to create a set of keywords that they thought were important and could be used to filter incoming messages. For example, you might have a keyword called “Knocksinna” because it was the code name for your current project. Each keyword was given a priority value with 1 being the highest. Each priority value was given an action chosen by the user. For example, “Place all priority 1 email in the Projects folder” or “Move all messages with priority 5 to the wastebasket”.

We then had a program that ran on the VAX/VMS minicomputer where ALL-IN-1 was installed. I believe that the program was written in LISP, a language that was often used for artificial intelligence (AI) projects. The program was written by my co-inventors, all of whom worked for one of DEC’s corporate software engineering group based in Sophia-Antipolis, a technology park located between Nice and Cannes in the South of France.

Artificial Intelligence was all the rage in 1987-88 and lots of work was being done in “knowledge engineering” to determine just how much intelligence could be incorporated into computer programs. DEC’s European Technical Center was staffed by some pretty interesting people, one of whom being Andrew Buchanan. He specialized in AI and convinced me that AI could solve some common business problems as opposed to the sample AI programs that were circulating. My favorite of these was one that allowed you to play Monopoly against the VAX. Unfortunately the AI wasn’t very developed, the computer was too stupid, or the human beings cheated a lot as the computer invariably lost by swapping valuable properties for a song. We viewed the Mail Filter as a project that illustrated how AI could assist people with common tasks in an office environment.

In any case, the LISP program took a list of new messages extracted from the user’s ALL-IN-1 inbox and compared the information about the new messages with the user’s keyword priority list before deciding how to process the messages. The program wasn’t fully integrated into ALL-IN-1 at the code level, so it wrote out a text file containing details of how the messages should be filed. An ALL-IN-1 script then opened the text file and executed instructions to process each message as instructed.

All of this sounds pretty kludgy and indeed it was. No great breakthrough in software engineering was achieved to create the various components that interacted to process incoming email, but that wasn’t the point. We wanted to demonstrate an idea and if the idea proved worthwhile, we wanted to take it forward and incorporate it as a new feature in a future version of ALL-IN-1. Unfortunately that never happened. The code that we created worked and although it was slow (and buggy at the start), it did the job. There’s no doubt in my mind that it would have been feasible to increase speed and eradicate any remaining bugs had we persisted but that never happened because other development priorities got in the way.

ALL-IN-1 never had inbox rules functionality but lots of email products boast a similar feature today. For example, the rules processing of Microsoft Outlook is simply a far more developed version of what we created in 1988. The large list of other patents that acknowledge patent 5377354 is evidence of its influence since the patent was filed by DEC.

Why was the Mail Filter patent never asserted in court if so many other email systems developed similar functionality? Perhaps it’s because the patent was just one of many owned by DEC, Compaq, and HP (for such was the ownership chain due to mergers and acquisitions) and never seemed to be important in terms of the overall patent portfolio. On the other hand, it’s true that DEC, Compaq, and HP all had patent cross-licensing agreements with other technology companies that allowed those companies to access the IP. Microsoft was one of those companies so any infringement by Outlook was fully covered by the cross-licensing agreement.

HTC bought the Mail Filter patent from HP in 2011. Companies quite often trim their patent portfolio to remove or sell on patents that have expired or no longer seem interesting and I’m sure that this is what happened when HP decided to sell the patent to HTC.

Looking back, we probably should have done more to develop the idea contained in the Mail Filter. At least we captured the idea in the patent.

Follow Tony @12Knocksinna

About Tony Redmond

Lead author for the Office 365 for IT Pros eBook and writer about all aspects of the Office 365 ecosystem.
This entry was posted in Email, Technology and tagged , , , . Bookmark the permalink.

2 Responses to The ALL-IN-1 Mail Filter: Forerunner of modern email rules processing

  1. Claire says:

    Interesting. To me, inter-network tasks are the most useful application of rules (eg if it’s going to rain, have my phone tell me to bring an umbrella to work). We’re building prioritization into our Exchange client, and originally thought about making it largely sentiment- and keyword-based, but it turned out that where senders lay on a recipient’s social graph was a much more useful indicator of relevancy.

  2. Art says:

    That was one of the first attempts to “commercialize” AI at DEC. This development led to rules integration becoming a critical feature for email clients. Several of the early products were sometimes promoted as having advanced AI-derived rules capabilities. I remember your TeamLinks and Banyan’s BeyondMail as two where that was the case.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.