Senator LONG. Doctor, don't you know it does not work quite that easily, though? Some of these overzealous agents and agencies go ahead and collect this information, and until some committee like this one comes along and checks into their activities, we do not know the things that they are doing.

Dr. KAYSEN. Senator Long, I know you are experienced with these matters, and your experience is great and mine is not. My previous experience in the Government impressed me with the power of the ap propriations process, and I thought that the appropriations committees and subcommittees can do quite a searching job of finding out what the executive agencies are up to.

Senator LONG. I have looked at these appropriations, and I have never found any appropriations for bugging the wiretapping equipment in many of the agencies-and they were able to have it and used it quite efficiently until this committee started looking into it.

Let me ask you one more question and I am through.

Let me ask you this: As I understand it, everything on an individual would be in there in regard to, let us say, Mr. Fensterwald. It would be accumulated in one group. It would be all in there. It would be the Fensterwald file?

Dr. KAYSEN. Yes.

Senator LONG. Would there be any time during the year when Mr. Fensterwald could go down and say, "I want to see what you boys have in my file."

Dr. KAYSEN. I want to say we honestly have not thought about that question very hard. The reason we have not thought about it hard is because the kind of information we are talking about does not raise the issue sharply. The information on Fensterwald will say he had 267 days of covered employment, you know, from his social security file, and so on. Then there will be 10 items from his tax return, and then it may be the report he made when his household fell into the onein-a-hundred-thousand households sample of the census, which says how many children he has and how big a house he lives in and whether he owns a refrigerator or not, and some other things like that.

All these things will be put together, and if, let us say-and this is the problem that I think you are addressing-if that file says quite incorrectly that Mr. Fensterwald owns a refrigerator and a freezer when, as a matter of fact, he only owns a refrigerator, and this would be because somebody mistranscribed something in a census form, my feeling would be, well, that is pretty unimportant. It does not do any harm to Mr. Fensterwald, and the statistical error it is going to put in is small.

Senator LONG. Well

Dr. KAYSEN. Excuse me, sir, can I continue?

Senator LONG. Go ahead.

Dr. KAYSEN. Now, if in this file, it started to contain more information, supposing it contains his efficiency rating, and it says, "Senator Long for whom he has worked says he is a very good employee, but Senator X, who isn't present this morning, dissents," and I think that would raise an entirely different question, Mr. Fensterwald would have a very different interest in being able to be sure that kind of information was correct if it were in the file.

Now, as I say, I think the reason that our committee did not give a great deal of attention to this question-we thought about it, but felt we could safely leave it for the moment-was we did not envision that the second kind of information would be in the file.

I think if the file starts expanding to cover that sort of information, this is a perfectly legitimate question which might be raised.

Senator LONG. Of course, Mr. Fensterwald, with the curiosity that he has, might be interested in looking in his file to see if they were accumulating other information or not, and he might not have any other way of knowing it other than taking a look at his file.

Dr. KAYSEN. Well, he might not, although again I seem to be able to summon up more admiration for the operations of the Congress than you gentlemen can, but my own reliance would be on the congressional control of the kind of information that goes in the file.

I am simply quite honestly unprepared to answer the question how feasible would it be to make it possible for Mr. Fensterwald to see the contents of his file.

We do not now do that, of course, in these separate agencies; that is, the Fensterwald file which now exists in dispersed form in the Social Security Administration, the Census Bureau, the IRS, and, if he owns a farm, the Department of Agriculture, and if he operates a little enterprise in the Bureau of Labor Statistics, and so on, that file exists. It is just scattered, and, as I understand the operations of the agencies, none of them provides the opportunity to Mr. Fensterwald to correct his file; although, of course, should the IRS engage in an administrative or legal proceeding, then the situation changes. It stops being a file merely and becomes subject to a different body of rules and traditions. Senator LONG. Doctor, you mentioned your belief in Congress. Let me say I have a very high regard for the ability and reaction of Congress. But, at the same time, I do not underestimate the ingenuity of various administrative agencies either.

Mr. Fensterwald, do you have any questions?

Mr. FENSTERWALD. I just have one comment and one question. Since we are talking about my security file, I think that it might be a great protection to the individual if he could have access to his file, even if necessary that he pay for retrieval of it, because you would get much less information pumped into the system, say, from informers-unevaluated information-if the agencies knew this would be available to the individual.

The question I had, Doctor, relates to the problem of whether this is wholly a future problem we have or does it have present implications? We have just totaled up the figure of $2,800 million individual files, which averages out to something like 14 different Federal agencies having files on an average American.

Is there any danger in all of this mass of information being put into a future Data Center or is this information in such incompatible form that we are only talking about a future problem?

Dr. KAYSEN. May I comment on your remark and then answer your question?

I agree with you about the possible usefulness of allowing an individual to look at his security file. I just want to repeat at every opportunity that I have that it is not our thought that security files or personnel files should be in the Data Center.

Now, on the question at to whether this is a current or future problem; I think it is mostly a future problem. At present, the data in each individual collection agency are so organized that matching is a difficult problem. It is not, however, impossible, and there are little experiments, exercises, in matching that have been going on, but they have all been on a sample basis; that is, you take a sample of returns of one sort and a sample of returns of another, and try to match them up. There has been no attempt that I am aware of to have wholesale matching, and the sample basis matchings show a great many difficulties. They suggest that unless you consider the matching problems fairly early in the way you design the whole data file, it is very difficult and expensive to do it.

Now, I think there is another comment to make and, perhaps, I am going beyond the question. At persent these large numbers of billions of files distributed over a variety of agencies are covered by quite a variety of disclosure and confidentiality standards, and one of the virtues we see in the suggestion that our committee has put forward is taking the occasion to try to get a uniform confidentiality and disclosure standard that covers all information that the Federal Government collects from individuals or, perhaps, one should say more narrowly all information that the Federal Government collects from individuals under compulsion.

It may be that there ought to be a special standard when a man cannot refuse to answer the question and, perhaps a different standard if it is a voluntary disclosure, although I think that is a question which needs exploration.

Mr. FENSTERWALD. I have nothing more.

Senator LONG. Mr. Kass.

Mr. KASS. Thank you, Mr. Chairman.

Dr. Kaysen, in your report you suggested that one of the major obstacles to the existing functioning of the systems is the decentralization in all of the various agency files. You have also suggested that the Bureau of the Budget should be responsible, in any event, for knowing what goes out, what type of questions they send out.

Yet, when we were working on the results of the survey that the Chairman sent out to all of the agencies, they were most interested in this type of information also.

Could not also one of the problems of this data bank be--not today, necessarily, but tomorrow-a huge decentralization that, as you say in your report, the growth of the problem will outstrip the strength of the remedies applied?

Dr. KAYSEN. Well, we conceive the data bank as really running in the other direction, trying to inject an element of centralization at one stage of the whole process. There is a collection stage, a storage and analysis stage, a tabulation and use stage. We would like to have the storage and analysis stage centralized. We are not talking about trying to centralize the collection stage. I think that could not be done, and it would not fit into the whole governmental system, really. We do not argue that there should be any attempt to centralize the tabulation and use stage.

We do say that, if the Data Center proves to be an efficient enterprise, other agencies of the Government which have statistical work to do, may ask the Data Center to do it for them, not by compulsion,

not by regulation, but simply because this is the most efficient way to get it done. Just as Government agencies now have appropriations for printing, and can print for themselves if they wish, but they often find it more efficient to get the GPO to print for them.

Mr. KASS. Will this be limited to other Government agencies? What about outside nongovernmental, State and local, medical institutions, educational institutions?

Dr. KAYSEN. In our report we suggested that the Data Center be set up with regulations, compensation arrangements, and capacity that make possible non-Federal government use. There would have to be a public purpose involved, and there would have to be a procedure which respected the confidentiality, and there would have to be, appropriate compensation. Within those safeguards, we think it would be wise to make this data available for State and local government use and for the use especially of research enterprises.

Mr. KAss. But is it a one-way street? In your report you seem to give the impression this is only a one-way street, that the data sent to information would be going out, but none of the information from State, local and other collectors would be coming in to the data bank.

Dr. KAYSEN. I am sorry if that is the impression we gave. That is not the suggestion we had in mind. In fact, we deliberately picked the label National Data Center rather than Federal Data Center to suggest that State and local information could also be put in.

We appreciate that there is really a problem of another level of organization and arrangements before you could do this. We had some familiarity with the Federal-level problems, and my own judgment is that we probably want to tackle the Federal level problem first, before we get into the more complicated problem of how State and local governments could also utilize this facility for efficiency and economy. But we certainly think that should be in the picture.

Mr. KASS. One more question, Dr. Kaysen. In your report_you suggested that a separate agency ought to be set up as a Federal Data Center, you call it. In your position outside of Government, have you given any thought as to whether this ought to be created as a part of the executive branch of the Government or maybe as a legislative arm of the Congress such as the General Accounting Offices?

Dr. KAYSEN. Well, in the report itself we made some recommendations about this agency. I have to say that we thought the arguments were strong for creating it in the executive branch.

For the agency to live and work successfully it has got to be in a cooperative relation, not a supervisory relation, to the statistical agencies in the executive branch.

The GAO, after all, has a legitimately-I do not want to say antagnostic, that is the wrong word, but adversary relation in the technical sense. It is looking over the shoulders of other people and saying, "Did you do right? Did you spend the funds according to the law?"

Mr. KASS. Maybe the correct analogy or a better analogy might be of the Library of Congress.

Dr. KAYSEN. Well, the analogy with the Library of Congress might be better I would have to say that I do not have a strong position on that. I think it is a suggestion worth considering.

The reason for the organizational recommendations we made is a belief that in order to function effectively the Center has to be in the hands of a group that things of running the Center as its primary task, not a secondary task after doing something else. On the other hand, the Census now has the biggest body of data and the biggest body of expertise, and we think in order to work effectively it has to be near to and in a close relation to the Census. That need not be precluded by your suggestion, and if you were to ask me offhand, and it would be an offhand answer, is there anything wrong with the proposition that you could take this whole apparatus and give it the same relation to the legislative branch that the Library of Congress has, I would say offhand I do not see why not. But it is not a question to which I have given prior thought.

Mr. KASS. Thank you, Dr. Kaysen.

Mr. Chairman, I have no more questions, but, there may be some additional questions that we would like to ask Dr. Kaysen, and we will put the question into a letter at a later date.

Senator LONG. Very well.

Mr. Waters.

Mr. WATERS. Dr. Kaysen, it is obvious that the interest of the subcommittee is in connection with the right of privacy dealing with the use of computers.

Don't you contemplate that before this type of information is programed, there is going to have to be a vast dredging up of records that were going to be evaluated and programed into apparatus of this type?

Dr. KAYSEN. Yes, there would have to be, as you say, a vast dredging up, although, as we have envisioned the way this would operate, we would try to start fairly currently, trying to get current material into the Data Center, and would go back only rather slowly. Going back in time is a very difficult thing to do since the older records were not collected with integration in mind, and I think that how much we could go back and dredge up is basically a practical problem of how difficult it would prove to be. My guess is that we really could not go very far back into the past to try to integrate past files very far back with great success, because of the practical difficulties.

Mr. WATERS. The information to be elicited from the computers would, of necessity, depend on the accuracy and authenticity of what goes in. I presume you have heard of a term which is new to me, but which is used in the programing business, they call it GIGO. Have you heard it?

Dr. KAYSEN. I am afraid I have not.

Mr. WATERS. It has been explained to me as meaning garbage in; garbage out.

Dr. KAYSEN. Yes, I am familiar with it.

Mr. WATERS. And it is a word they use in the computer business, and the people who program them are familiar with it.

We hear from time to time about individuals who, perhaps, may receive a bill from the department store or a charge account which they feel they are not liable for.

Dr. KAYSEN. Right.