Doug Engelbart's    Colloquium at Stanford

An In-Depth Look at "The Unfinished Revolution"

I work for Microsystems. I am delighted to be here today. Partly because this is the first presentation that I have given in forty years that isnÂ't about S&L. I said during one of the sessions here, a proposal had been made for the beginnings of doing a dynamic knowledge depository group. I said hey, I have an application in mind for this. Doug and Marcel had been kind enough to give me a soapbox on which to talk about this. The seed for this  idea that I am about to spin out for you is by a fellow Ken Klemmens of Foresight Institute, a senior associate. We met at a nano smooze. I had come  from a trying day of a session of an organization that I work for called Oasis and the United Nations. We had assembled a hundred and fifty of the top  electronic people in San Jose. I had staggered out of this meeting saying give me something else. I sat down with Ken. Here I had spent several years in  what you would consider an improvement community. In W3C and this other one in OASIS and the UN, developing this thing called S&L. Ken had  spent a similar amount of time in this organization called IEEE developing 802.11, an incredibly complex application of wireless networking. So, we were  comparing war stories about this. The idea of how you actually get standards to work and people to agree on things. He came up with this idea, which  you are about to see the result of. I went home, and it kind of cooked. I give Ken the credit of what came out good in this and take the blame for the  more bizarre aspects of what I am going to suggest. I am chiefly interested in the idea of collective IQ. I have seen it work, it works, and itÂ's a wonderful  thing. Groups of people can come together and come up with things that are beyond one person. That is my idea of collective IQ. Doug has pointed out  in this colloquium that we have a number of problems. We are going to use collective IQ to solve this problem. I believe that is true and it has an effect  on all of us. Solutions that have been proposed revolve around the idea of better information. I think that better information is necessary, but I would  like to start with my observations about that. My observation is not all disagreements are due to misunderstanding and lack of information. Probably  the most obvious example is the abortion debate. People disagree with each other on this subject. They are not going to stop disagreeing if you provide  better information to them. A social policy problem that is not going to be solved by better information and more communication. My observation is  that a lot of the problems share this property to a large extent. We spend a great time earlier about the energy problem. Anyone who listen to the man  who explained the size of the energy problem and then went off and though about it for a few minutes, I think would come to the concluding that we are  going to solve the energy problem by using less of it. How are we going to do with less energy? There are two ways to go about it. Are we going to  share what is left fairly or will we let the rich people use up what is left? 
 

So, I mean the basic question are we or are we not going to get people out of driving big cars and on to the bus or not? That is not a question that is going to be solved completely by information alone. So, I would like to talk about augmenting the DKR itself. Dynamic knowledge repositories  are essential to what I am going to propose, but I would like to say is that they are a necessary but not a sufficient part of the solution. The big  problems are not going to just require us to design solutions but we also must agree to be bound to the solutions. We are gong to have to be able to  take someone who will not get on a bus and drag them on. If we are going to solve the energy problem. It is not enough to say that now you know this  is the answer. We are going to have to compel that. How do we facilitate the process of making legally binding decisions? How do we do that? We have  a traditional answer to this problem. We have a process. I brought some examples. Here are the standing orders of the house of commons of parliament.  This will solve large problems. This is not new, pieces of this go back four centuries. ItÂ's called the parliamentary process. In the U.S., we have a  standard form of this thing. ItÂ's RobertÂ's Rules of Order. 
 

How many people are aware of the RobertÂ's Rules of Order. How many people know about RobertÂ's Rules of Order, I didnÂ't say love it. This is how Ken and I got started we both happen to be working for organizations that are run by RobertÂ's Rules of Order. If you go look, you will see that this  thing is deeply woven into our legal and social structure. Pretty much every corporation, university, all 87,000 governments are run based on this or  some interpretation of this. So, we have a process. What is wrong with it? It is slow; it is complicated and worst of all it does not work on-line. I will not  get into it. We donÂ't want to deal with this. Instead, we go to consensus based solutions. We will just develop a consensus and do this. The problem is  that consensus based processes wonÂ't deal with the problems when people just do not want to agree on things. I donÂ't want to give you my  consensus, what are you going to do about it? Consensus is not going to work for a lot of these problems. If only our parliamentary process worked  on-line. There are some really good things about this ancient thing. It is the epitome of democracy. When we say democracy in practice, we are talking  about RobertÂ's Rules of Order. That is what majority rules this amounts to when you work out all of the corner cases. 
 

ItÂ's comprehensive, itÂ's fully documented. You can go into any bookstore in the country, and get a copy of Roberts. You can order it on amazon.com. It is thoroughly debugged. The reason that it is seven hundred pages is that they finally figured out all of the hard problems. There is a  solution in there somewhere. Surprisingly, it is capable of bootstrapping itself. You can take a RobertÂ's Rules of Order. It starts with a group of people  milling about in a room and turns it into what is called a deliberate of assembly. That is pretty amazing. 
 

The heart of the problem is that mail is too slow to conduct a real parliament process. You can conduct pieces of it. You can hold mail ballots, but you cannot amend motions. The cycle is too long here. The solution is not to simply transpose the traditional process isomorphicly into e-mail. We  need a different way of looking at this. I will suggest to you a different way. My observation is as follows. Any process that is set up according to  RobertÂ's is a state machine. I 
 

A traditional parliamentary process constitutes a state machine. I am not saying it could be, it is. What is its state? It is the information that gets saved when we adjourn. So, if you say at any moment in a deliberate of assembly, we are adjourned, you move from state to state. We move from state  to state in a determined way. The parliamentary motions (84) can be seen as commands to the machine, the instruction set. Perhaps we can instantiated  such machines in software. Interesting idea, where does it go? 
 

Ken and I came up with this thing called the Parliamentary Assistant. LetÂ's think about one of these machines running on a web server. The server takes care of all of the procedural details, and maintains the document base, so it is doing a great deal of work here. Interaction with this machine  takes place through the forms 

that the server generates to the user. Not unlike ordering a book at amazon.com or working with a game. If we can make this thing work fast enough and interactively enough, the social dynamic of such a thing could start to resemble the social dynamics of a multi user game. Perhaps we could take  technology that we are using for electronic commerce and technology that we are using for D&D and put it together to actually make a framework by  which you could actually make decisions. 
 

Here are some of the benefits of this thing. First, all the procedure is handled by the server. All seven hundred pages of impossible to follow nits can be handled by the machine. One of the consequences of this, I know all of you have had experience of a RobertÂ's process where some jerk got into  the process and started raising points of order. You canÂ't have a point of order; you canÂ't do it because you are not offered any choices that are not  legal. There goes that whole bunch of stuff that is no longer a problem. You canÂ't have priority conflicts between speakers. Most of the machinery of  traditional parliamentary procedure is saying, " You have the floor" no, " you have the floor". ItÂ's about people not interrupting each other, but in this  kind of a setting, you canÂ't have people interrupt each other. Something comes in one millisecond later, and it is later. It is not a problem deciding what  the prorate is. The key to this if it works, is if it is properly implemented it can be substituted for exiting processes. I am not talking about something that  would be cool in helping us out. I am talking about taking the school board down the street, or General Motors Corporation, or the state of California, or  the Bishops of the Lutheran Church, or any of those organizations that is run by RobertÂ's or something close to it. Tomorrow you can start using this environment that is fun. And a lawyer says that it is legally equivalent to what you were doing before. If we canÂ't do that the idea doesnÂ't work, so that is the idea. 
 

If you start thinking about the Parliamentary process as a machine, you can start applying some of the concepts of a machine. For example, the compete of recursion. The traditional process is saturated with the concept of computer recursion. When you amend something, you make a motion to  amend, and then you make a motion to amend the amendment. At each stage, you are calling the same process. In fact, RobertÂ's says the language they  use is said to reoccur. A related concept is the idea something multitasking. Which is another layer of recursion in the sense when People are getting  together to do this process are getting together to form a committee, they are spawning a child process. They are creating an automaton. In  implementing this thing you are getting a lot of efficiency that we say once we have the machinery handling a motion, or any stack of motions once we  have the machinery for setting up one of these things, we have the machinery for setting up any level of subcommittees. Optimization. We can start  thinking about machine level optimization. One of the things that really slows us down is that you can only talk about one thing at once. For example if  a motion is before us and at the same time, you want to talk about shall we stop debate on the motion? We have to stop debate on the motion to talk  about if we should stop debate on the motion. Because the motion to stop debate on the motion is not itself debatable, there is no reason you have to  do that. In a machine, you could have the motion to stop the debate in the background, on another track from the thing you were talking about. You  could start collapsing some of the time frame. You could extend the thought to say letÂ's separate the main motion and itÂ's amendment to discussions  about when we will hold the next meeting or shall we remove somebody, things that are administrative. RobertÂ's actually makes this distinction in  language that you have to work really hard at to figure out that that is what is going on. 
 

We could have user interface. We could use current user interface engineering to create a different kind of environment. All of the machinery that we are currently using to do business can be used for this purpose. Fill-in forms, menus, this kind of thing. And notice that the form that is  generated for any user at any instance in this process has only the legal options available for that moment. You donÂ't have to remember what it is that  you have to do. It is like ordering, the only to be presented with things that you are able to do at that point. I think a really slick interface to that kind of  thing, would not only show you what is legal to do, but it would show you a preview the state of the machine if you did it. Basically, what you could  do, the thing Ken and I were observing is that Roberts works really good once you understand it, once you have studied up on it. In fact, if that were  not true, we would have labor uniting and church groups, and sewing circles, and popular forms of this thing work. They way that they do it is appoint  one of their people to be an expert and call that person the parliamentarian. What we are saying is that you donÂ't have to do that anymore. This process  will hold you hand. One of the best things I like about this is the remaining function of the human chair. 
 

It is the chair that slows this ordinary process down. If the chair doesnÂ't have to decide for every conflicts of who has the floor, if the chair doesnÂ't have to declare what the procedure is and if the chair doesnÂ't have to doing a bunch of things it does. You have cut a lot of latency out of the  operation of the machine. However there are going to be some things that are real problems. Setting agendas, I donÂ't know how to get around that one.  Checking lexical form. RobertÂ's says that you cannot say I move that we do not move this. You have to say I move we do X. That is a classic example of  something that is brain dead simple for a human being to decide, and really hard for a machine. The problem that what is relevant to a discussion this is  a problem. I am not trying to say that this has completely been thought through I am saying that more work needs to be done. 

If this could be done, it will have potentially revolutionary implications for society. 
 

I think that the distributed yet synchronous but legally binding proceeds would make possible global decision making by people who are distributed in space and time. Local decision-making. I would like my neighborhood to decide what the speed limit is going to be on my street. And  everything in between. All 87,000 thousand of those governments can be implemented this way. If it works, that remains to be seen. This could be a  viable way to solve the big problem we are facing. There is a lot of promise here and a lot of working out to do. This is the first group I have shown this  idea to. I think that it is directly related to the idea that we are talking about which is the improvement of process and decision making. It is going to take  some work to find out if there is going to be anything to it. The first reaction I had from someone was a very smart person that I work with at SUNN. He  could not decide if this was insanely great or greatly insane. I donÂ't know either and it is going to take some groups to figure that out. 
 

The relationship here is that these parliamentary assistants assume the existence of a dynamic knowledge repository. This is an AP for a DKR. You have to be able to handle large bodies of data and cruise around in them. And say where were we last week. That all has to be in place before this  idea can work. There is another thing, like in this course, where we are in the layers of abstraction. We have improving things and we have the process  of improving the improvement process. If this is practical, this would be a way to do the kind of work that Ken and I do. Which is where we came from in  the first place. Gee wouldnÂ't it be great if in our committee work, we had something like this. This could provide governance as well as being an  improvement process. The development of this idea, if I am able to find enough crazy people. We will work with in OASIS. (Organization for  Advancement of Structured Infrastructure Standards) It is easy to join, under two fifty a year, and it is using Roberts. So, we will use the process itself  to develop the process that we are using. There is a certain neatness that appeals to geeks about that. 
 

If you are interested in this, you can look at two URLs that are on the screen. One is the proposal itself, which is in much more detail than what I have just said and a preliminary plan of action. Both URLs begin with the string http://metalab.unc.edu/bosak/pa. The only difference is one is /pa.htm\  and the other one is /pa-act.htm. You can also contact us by mail with the address at the bottom of the screen. This slide will be on-line. Basically, when  you read these plans of actions, Ken and I will be checking around whom else thinks that this is to put some time into. If you think it is interesting  enough after reading these two documents, then get back to us. Thanks.

[  principal lecture]

 ---