the data together and making a forecast-there is no public interest in disclosing what Mr. Long of Missouri does on his farm.

Senator LONG. But they have on this, and this is confidential, and it will go no further

Dr. KAYSEN. Yes.

Senator LONG (continuing). Which I doubt, but if they put this into this data center, then this Internal Revenue agent (who, incidentally, had to come down to my farm and count the cattle some time back; we had several hundred-he decided it was a business operation and not a plaything) he would have to go to this Department and say, "I am a certain agent and have to find out how many cattle this farmer has and what he is going to do with them.'

Dr. KAYSEN. No, sir.

Senator LONG. Why couldn't he do that?

Dr. KAYSEN. As our proposal stands, and as I would recommend strongly, the Internal Revenue agent cannot do that, just as data from other Government agencies go into the census today, but no individual data ever comes out of the census; even a committee of Congress does not get data on an individual return.

The Internal Revenue might ask that Data Center for a tabulation saying how many farmers in Missouri have over a head of cattle on their farms, and how many last year, and a tabulation will be given them.

Senator LONG. As I understand, this agent or the Internal Revenue Service could not come in and say to the Department, "My number is so and so and I want Tom Smith's income. I want to see how many cattle he reported out on his farm."

Dr. KAYSEN. No.

Senator LONG. What would keep him from doing it? You said they have to have the name and the social security number to identify it.

Dr. KAYSEN. What would keep him from doing it would be, as I hope, a statute which said, "Here are the rules under which the Data Center collects data and here are the rules under which it gives out information," and I am suggesting that should be similar to the one that governs the census, and that forbids disclosure even to other agencies of the Government of information on an individual.

Senator LONG. But they do get that information out, and since you are identified in this computer by your name or social security number, an overzealous agent, it appears to me, could get that same information. Dr. KAYSEN. I think we are back to the earlier point, Senator. I must agree with you that we cannot prevent violations of the law, but we can protect against them by making them difficult, and we can cut down their frequency by the administrative arrangements we make which protect things.

I cannot say this will never happen, but I think we can design an enterprise in which it won't happen very often.

Senator LONG. Senator Thurmond.

Senator THURMOND. Thank you, Mr. Chairman.

The Federal Government today is requiring people to fill out so many blanks, so many forms, that some businessmen tell me they are just so discouraged they feel like going out of business, and some of these information forms they fill out require information to be furnished that seems to have nothing to do with the subject at hand.

Of course, you are not familiar, I guess, with all those blanks I am talking about. But, for instance, some of these blanks have asked some very personal questions.

I want to ask you about this: Over at the University of Maryland, they gave out the grades on a computer system, and they got all mixed up, and the students got the wrong grades, and they had a lot of them flunking who passed, and they were about to destroy the reputation and the records in the college of a lot of the students.

What do you have in mind to protect the integrity of such a system as this if the Government should adopt some system of this kind?

Dr. KAYSEN. Let me say, Senator Thurmond, that this is not a field in which I would qualify as an expert, and I am talking as not the best witness, but I will answer the question. I am not trying to evade it.

I think that there are many ill-designed computer systems that are at work in the world today. I think there are some very well-designed computer systems that are at work in the world today. I think there are some very well-designed computer systems. In the Government there is a great deal of experience and expertise on the design of computer systems and the use of computers. I would expect that the Government would put in the effort and the money, and it takes money, to have the kind of system which has built-in checks, and we can design a system with built-in checks against the sort of thing you mentioned.

I do not know anything specific about the University of Maryland incident and, therefore, I do not want to say that it is the result of a poorly designed computer system. Maybe it is and maybe it is not. But there have been examples simply of poorly designed systems, illthought out programs, in which these things have not been taken into

account.

Many at the Federal statistical agencies have already a great deal of experience with the use of large-scale machine processing installations, and they can draw on that experience and draw on outside technical expertise as well, and look to avoid those mistakes.

Senator THURMOND. I can see where an individual might be hurt badly if the wrong information should be furnished to an appropriate source, if it is inaccurate and incorrect. It might hurt him tremendously, just like these grades the University of Maryland gave out, which finally had to be corrected. But they were about to ruin a large number of the college students with their records.

It seems to me if some system as this is ever adopted, there must be some guarantee or some kind of an insurance that the system is going to be thoroughly tested and proved to be thoroughly accurate before it will be used in operations.

Dr. KAYSEN. Senator Thurmond, I think that is important. But I think more important is the guarantee that this system won't have as its aim or purpose to furnish information on individuals. It will have as its aim to furnish statistical information on groups of individuals.

Now, if you make an error in an individual report on the basis of which some action is taken, a tax return or a personnel report, an administrative report, then individuals may be grieviously harmed, just as in the example you mentioned.

But, as the Data Center is envisaged, it would be up to the agencies who have the responsibility for using the information they collect, to use information on individuals. When the IRS wanted to proceed, let us say, against Mr. X because they feel that he has not properly filled out his tax return, they would act on the basis of what is in their files. The Data Center would not be in the business of dealing with individuals and trying to collect taxes from them. It would be in the business of taking some of the information that gets into tax returns, and assembling it together with other information and trying to get a statistical picture, not a picture of individuals.

I would say the most important safeguard is the distinction between administrative agencies which, by their nature, have to deal with individuals, and the Data Center which is essentially a statistical enterprise which deals with groups and not with individuals.

Senator THURMOND. Would you envisage a Data Center of this kind to be useful to one Government agency that is considering employing a man who is already holding a position with another Government agency, and they may wish to get his records from this Data Center? Dr. KAYSEN. No, sir; we would not.

In our conception we would say all dossier-type information, personnel files and that sort of thing, should not be in this Data Center. Now, if the Congress

Senator THURMOND. This is to apply to groups only then?

Dr. KAYSEN. It is to apply to the kind of information which we want to assemble for groups of people or households or business enterprises or farms. It is not to contain the kind of information that says, "Mr. X drinks a lot, he has been divorced, and his wife, ex-wife, has the following complaints against him," and so on, which might bear reasonably on the question of whether he is suitable for such and such employment.

1

But really, that is not the kind of information which anybody connected with this enterprise envisages putting in a data center, and I would best describe the information that we do envisage by talking about what is collected by the major statistical agencies, which we have mentioned several times, rather than the kind of thing you mentioned, which would come out of a personnel employment file or something of that sort.

Senator THURMOND. As you envisage this then, I guess the integrity of this system would depend upon accuracy and the data being placed in it, and the accuracy of those who distributed the data. In other words, it must be accurate to go in and it must be accurate to go out. Dr. KAYSEN. It must be accurate to go in, and it must not come out for any individual. It must come out only as group data of some sort or a frequency distribution.

You know, there are so many people who earn an income of so much and higher, and so many who earn an income from here to here, and they live here and they live there, that kind of data. But no data which says that Mr. John Jones lives at such and such an address and he earns such and such an income. That would not come out of the data center. It might go into the data center from the IRS. It would not come out.

Senator THURMOND. Who controls what goes in and who controls what comes out would really determine the efficiency and usefulness of a center of this kind, would it not?

Dr. KAYSEN. That is right, sir. And as we envision it, there should be a general legislative statement about what kinds of things go in. I do not think it is in the nature of a legislative enactment that it could specify every last item that goes in. That seems impossible.

But I think there should be a legislative policy statement that says the data center is meant to hold such and such and such and such kinds of information, and then, as I said in my statement, I think that the data center, the Census Bureau, the various contributory agencies that each year go through the appropriations process, will have to face up to the question asked in the subcommittee, "Have you added some new information to your form? What is it? Is that information which is going to the data center? Why? Why do you put it there? What is the use of it?"

This is essentially the procedure that the census goes through now when it adds a new questionnaire or adds a new census form. It discusses its plans with the Appropriation Subcommittee, Independent Offices, I believe it is, and-no, it is Commerce, is it not-and there is always some congressional control over the question "Do you add a new form? Do you add a new question or questions? What are they?" And if, as you gentlemen know better than I, if one slips by, if something is done, then there is retrospective control, and the committee the next year can say, "How come you did this? Why did you do that?" We are suggesting that this process which has worked with great success for the Census Bureau be adapted to the Data Center. And again, if I am repeating myself, Senator Long, excuse me, but I would emphasize the great importance of a statutory restriction which says that this enterprise may not disclose information on an individual respondent to anybody, including other agencies of the government.

Senator THURMOND. It might be wise to have a heavy penalty that is applied, too, because, as Senator Long has brought out, evidently someone with the Internal Revenue Service did reveal this confidential information about this particular taxpayer.

Dr. KAYSEN. Yes.

Sir, may I make a comment on a point that Senator Thurmond raised because it has not come up yet, and I think it is relevant. You talk about the burden on respondents of increasing questionnaires, and you were concerned about all the paper the Agriculture Department is sending you.

One of the purposes of a data center that my colleagues and I had in mind was precisely to minimize that burden, to make sure that where information already exists in the Government, where it is collected by one agency, another agency does not have to go out and collect the information all over again, especially if we are talking about information that is not used for administrative purposes.

It is obvious that if you are enforcing, let us say, a licensing regulation you have to get the data on individuals. But if you are doing something like trying to estimate cattle production you are not interested in the data on individuals, and you want to be sure that all the

information that the Government already has is utilized before a request is made for new information.

Senator THURMOND. And you feel this Center would be of benefit along that line in helping to eliminate some of the blanks and forms that business people and others have to fill out?

Dr. KAYSEN. I would like to put it a little more cautiously, Senator Thurmond. I do not want to promise it will eliminate them. I think it will slow down the growth. I think we have experienced the growth, and the growth responds to a demand for more information which the Government feels it needs.

If one looks ahead, one sees this demand increasing, and I think if you see how in the face of that increasing demand you prevent people from being swamped with questionnaires, then I think you can say you're making sure that you make the best use of all the information you already have.

Senator THURMOND. Of course, I think the public and the taxpayers might as well recognize the fact that when the Government continues to go more and more into every facet of people's lives, literally every phase of activity, which was never envisioned by those who wrote the Constitution, that there is going to be more and more requests for information because the more you approach the welfare state, the more information the bureaucrats have to have to make their reports. So it is conceivable where there will have to be more redtape and more records than there used to be when the Constitution was followed.

Dr. KAYSEN. Sir, that is a policy matter in which, perhaps, we hold different views.

Senator THURMOND. Thank you, Mr. Chairman.
Senator LONG. Thank you, Senator.

Let me ask you another question or two, Doctor.

Do you see any danger or do you see any possibility that some of our overzealous agencies, including, perhaps, what Senator Thurmond suggested, might feel that this is just getting their foot in the door. While you are using this now just for the purpose of gathering statistics, and will put that information out on a group basis, that some of the overzealous planners might try to include everything from the cradle to the grave as time goes on?

Dr. KAYSEN. I think there is some such danger. I think there is a fairly natural drive on the part of people who are running an operation to make it bigger. I have to admit that-I won't admit it about myself, I will be more charitable and admit it about my brethren and say that social scientists seem to have sometimes an insatiable curiosity, wanting to know more and more. But I am fairly confident that the checks which the governmental system imposes on the action of the executive agencies are strong enough to control it.

It seems to me that the control of this enterprise will always rest with the Congress. The Congress is always in a position to say, “No, we don't want to see you collecting information X, Y, Z, and if you put that on your questionnaire we are not going to appropriate a dollar for you to do it," and if that control did not exist, I think the dangers you are talking about would be more serious.