The RAND headquarters building at 1700 Main Street in Santa Monica, California, was designed by John D. Williams, head of the Mathematics Division. He raised the idea of creating a building that would increase the probability of chance personal meetings. In a December 26, 1950, memo to RAND staff, Williams argued that such meetings would promote the interdisciplinary aspect of RAND — that is, the use of mixed teams of analysts to address a problem.
Original document accession number M-4251.
From: J.D. Williams
26 December 1950
To: RAND Staff
Subject: Comments on RAND Building Program
These notes are going directly from dictation to vellum, so as to save a little time; it follows that I am not prepared to defend their organization, the grammar, or anything else about it that looks weak in the light of day.
We suddenly find ourselves in a rather advanced state on our building program. It has changed from a vague gleam in our eyes, which was the steady state for several years, to a state where the building site is looking pretty firm and the architect is about to walk in the door. It's a little like Korea: our preparations for the event, while far from negligible, leave something to be desired.
We have discussed this subject among ourselves only a little, and usually at times when we felt we should be doing something else. We have had time to sound off extreme ideas, ranging from an eight-story obscenity at Hollywood and Vine (with a supermarket on the first floor), to whatever the character had in mind who wanted to be able to hear a cricket. I feel that the subject deserves some thought and time, that we should try to reach a meeting of minds, and that now is the time to do it.
If it does not more than provide a basis of disagreement, I thought it would be valuable for somebody to set down some of the facts and fancies that we hold. With these as a starter, I hope we will add and subtract items until we arrive at a set of essentials which cannot be compromised, at a set of desirable entities, and a set of things to be avoided. This memo is not organized along those lines.
While we undoubtedly require several office types for single occupancy as well as several for multiple occupancy, it is fairly important that the desirability of offices, within a type, be fairly constant throughout the building. Otherwise those who get the less desirable ones will be unhappy.
I believe that almost all of our requirements can be satisfied, and in a reasonably priced structure, if we have enough land. Since we have a strong internal reason for building compactly, the argument in favor of a substantial piece of land is not that it is desirable to spread out the building. The principal function of land is to provide insulation and isolation, and spaciousness. Insulation from noise, isolation from distractions and from the public gaze — my present office windows are high to begin with and then the bottom section is frosted — just so that people waiting for buses, or otherwise loitering, cannot participate in my meetings.
One of the really interesting, and pressing, questions is: Can we put up a building on a piece of land as small as that opposite the City Hall and still get approximately what we want? On my own part, there are times when I think we just can, and other times when I think we just can't.
Just to have something concrete to think about, I invented some typical offices and structure to go with it, and have been trying to fit on to that piece of property. For purposes of illustration, I will describe them briefly; but let me say that I appreciate that others will have ideas, and doubtless better ideas on sizes and arrangements — I am only prepared to shed blood from my veins in defense of these; not artery blood.
But first, recall that RAND now has about four hundred and fifty people, of whom about three hundred are office dwellers; the rest live in Washington, machine rooms, and open areas, or they are nomadic, such as the guards and janitors. These people live in about two hundred offices; about one hundred of these are singles; about seventy are doubles — correction, about one hundred and ten of these are singles — about seventy are doubles; and about twenty are occupied by three or four persons. I guess that in a new building there would be fewer of the last and more doubles.
I consider three office sizes, for purposes of exploration. These were — these are — nine by six and one-half, nine by thirteen, and eighteen and one half by thirteen; inside dimensions, in all cases, I require that the first have one window and that the other two have two windows. The first is suited only to single occupancy, the others, single or alterable or for two occupants.
I hypothesized six-foot width for hallways. Offices of these types can be accommodated in a building element, or wing, or section, thirty-four feet wide, outside measure, allowing six inches for each wall. The hall can be central with 9 x 13 or 9 x 61 1/2 offices on both sides, and the hall would be at one side where 18 1/2 x 13 offices on both sides, and the hall would be at one side where 18 1/2 x 13 offices prevail. I have played with this a little, and it is pretty flexible and efficient. This model at least serves to give some ideas about total building footage required.
The first peek at the problem suggest that our present size of organization and the size of the urban properties that are being considered, a one-story structure is infeasible. This is strengthened if we take into account the possibility that RAND may grow. I have, therefore, considered the elements of RAND in two categories, research and non-research, in the hope that at least the research people could be kept conveniently together on one floor. This does not quite get us off the hook; however, because future growth of RAND, if it occurs, would presumably be growth of the research arm principally. Therefore, the building plan should be flexible enough to encompass (say) another 150 research people.
The 9 x 13 offices mentioned earlier are a convenient unit. In terms of them, I estimate that our present needs for research offices and associated facilities (e.g., conference rooms) are 200 units. Similarly, the non-research functions (e.g., numerical analysis, publications, business administration, etc.) seem to require about 150 units. In the future the research elements might require 300 units and the non-research perhaps 200 units. If you treat the 9 x 13 unit as occupying 13 x 13 1/2 of building space (thus including the walls and half the hall), you get 175 square feet per unit. This indicates that the present needs for research and non-research are like 35,000 and 25,000 square feet, respectively, and that these may eventually change to 50,0000 and 35,000, respectively. Using a standard building section, 26 feet wide, the research staff would need about 1350 linear feet of that structure and the non-research staff would need about 1000 linear feet of such structure; eventually these might change to 1900 and 1300.
Now having some idea about how much space we needed, and of a way of packing it into a building section in a manner to provide light and air for all concerned, I was then ready to festoon the landscape with this stuff in some useful manner, if possible.
Judging from modern buildings I have seen, a popular way of spreading stuff around is to arrange it like the bone structure of a fish: a central structure with wings jutting out at intervals on both sides of the skeleton. It may be a useful design for us, if we have enough land. But for the parcel which we are now considering so seriously, the fish skeleton eats up the available space very fast; and it is not the best design for inter-office travel.
I was therefore led to try a system of closed courts or patios, and became involved in the theory of regular lattices, which is a fascinating subject; the square, the figure eight, and a hierarchy of more complicated designs.
As one rough measure of the utility of such designs, from our special point of view, I have taken the average distance between offices, measured along the grid, i.e., the halls. With any intelligent arrangement of our people, we can do better in practice than these average distances, but they still offer some information which is pertinent to us. I insert here a little table which gives the average distances for various simple lattices, expressed as a percentage of the total length of all the halls in the building. I also exhibit a picture of one of the lattices. The last column (S) in the table shows the overall size of the lattice. We note that the average travel distance decreases as the lattice becomes complex. that the decrease is most noticeable for the square lattices, and that the decreases are particularly noteworthy in the first two or three steps. For example, in a nine-patio lattice (i.e., a 4 x 4), the average distance is 10 percent of the total length of halls, as compared with 25 percent in the case of the simple square. Also, the size of the building as measured by its side S is half as great as that for the square.
Average Distance Between Points in Lattice (measured along lattice lines expressed as a percentage of the total length of the all lattice lines)
|No. of vertical lines
|No. of horizontal lines
S = length of the square lattices (as percentage of all lattice lines)
Of course, you have to watch it a little when you begin to translate the lattice into building structure. The building sections have finite width (twenty-six feet in my model), which reduces the width of the lattice openings; the people who live there, and who were so happy initially, will begin to notice it if the patios shrink too much. What the useful minimum is, perhaps our architect can tell; perhaps our psychologists can assist him.
A little playing with this will convince you that it is possible to put us on a fairly small piece of land — maybe. For example even our hypothetical future research organization, requiring nineteen hundred feet of halls on this model, could be fitted on and around a nine patio system, measuring two hundred and sixty-four feet on a side; on the three hundred and forty-four feet deep property opposite the courthouse, this would permit a forty foot set-back on front and rear. The patios would be like fifty-three feet across. I don't know what you would do with the non-research fractions of RAND. Of course, a second story would put over two-thirds of the ground floor. Such second story space might be regarded as less desirable than the first story space, and it might depreciate the first story space: for example, the patios might have to be larger to be equally desirable what with all that over-hang (correction — what with all that structure rearing up about it).
There is another way to make the patios somewhat larger, and which would make small ones more tolerable — and which would moreover reduce the average interoffice distance below that of the lattice: it might be that, in view of climatic conditions here, we could throw all or most of the halls out of the building. The patios could be surrounded by porches onto which the office doors would open. The porches would provide cover against the rain on those three or four happy days each year: otherwise, one could cross the patios from office to office. These small sheltered areas would not be windy.
My guess is that if a multiple patio scheme were artfully done, it would develop that "outside offices", in the normal sense would rank low in popularity for they wouldn't be less quiet and intimate; in fact, the vista from them would be filled with unrewarding objects such as automobile traffic and the rear of the Elks Club. Of course, the number of outside offices diminishes as the lattice becomes more complex.
Later: After seeing the above, I was tempted to fix it up a little. But reason prevailed: the way it is, those who disagree with a point will have to guess whether its me they disagree with, or my secretary, who often construes my "would"'s as "wouldn't"s, etc; conversely, they may find me biting their flank just when we seem to have reached harmony.