In the discussion above, we assumed the existence of a great deal of technological infrastructure--the existence of computers and networks of computers of various kinds, and of certain types of software. If these weren't available, none of us would be able to send email messages. There is also, of course, the existence and well-functioning of the larger electrical systems within which the computers live. The existence of this infrastructure can be said, in turn, to be an enabling condition of all our computing activities. This is quite typical: something the agent does is enabling only in circumstances in which enabling conditions have been established by the society, that is by the culture and technology the society has produced.
In the modern world, the things an agent can do are most often not merely a product of the movements the agent can execute, but these combined with various artifacts: structure and equipment provided by human beings. Broadly speaking, one can distinguish between the infrastructure at a location, and the equipment that an agent has. When we drive or bike to work, for example, the roads from Palo Alto to Stanford are parts of the infrastructure provided at these locations. The cars and bikes we use are part of the agent's equipment.
When infrastructure is provided at a location in order to make it possible for agents to achieve certain goals, there is a presupposition about what the agents can do, the abilities they bring with them to the infrastructure. A stairway, for example, seen as providing a way of moving from one level to another in a building, presupposes the ability (roughly) to lift oneself eight inches. This is most naturally done with the legs, but can be done with the arms, and also with some very high-tech wheel-chairs. A ramp, provided for the same purpose, only presupposes the ability to move forward, something that can be done by walking or by using an ordinary wheelchair.
The presupposed abilities are the ones that are necessary to interface with the infrastructure, to make use of it for the purpose in question.
Consider again the case where it is essential that people be able to move between floors in a building--typically as an act enabling further, doubtless more important accomplishments. Suppose the architects made a bizarre mistake and forgot to put in stairs. No one except dedicated wall-scalers can get from one floor to another. Then stairs are installed between floors. This changes the circumstances at the location and this change enables anyone who can climb steps to move between floors. Almost anyone who can walk can climb steps, though the bipedal movement types involved are actually different. And, of course, we are assuming the stairs are well-designed to enable most people to use them. Unless the stairs are very wide and shallow, though, it will be hard for people walking with the aid of crutches to use the stairs. Further, some who can't walk can use stairs as well, for example if they can crawl on all fours and lift their upper bodies sufficiently. So some people will walk; some will crawl; who knows, perhaps, some will walk on their hands--all, in their different modes, are now able to move between floors. The stairs provide enabling conditions in which all of them are able to do something they couldn't do before, at least not in the preexisting circumstances at that location.
Still, some are left out; those who can't both move forward and lift their bodies sufficiently. Let us call what the walkers, crutch-assisted or not, and the crawlers can do moving point to point and lifting; perhaps there are others who can can only move point to point. Both of these are accomplishments, not executions; they are accomplishments which can result from executions of many different movement types. Given (normally designed) stairs at a location and relative to the desired accomplishment of moving between floors, moving point to point and lifting is what we shall call an interfacing accomplishment. In those same circumstances and relative to that desired accomplishment, simply moving point to point is not an interfacing accomplishment. In building the stairs, the engineers can be thought of as changing the circumstances at a location in a way that allows the extension of previously existing way-of relations, at that location. Anyone who had a way of moving point to point and lifting, now has a way of moving between floors at that location. Thus it makes sense to build stairs because we can assume that the majority of people who need to move between floors will, independently of the stairs, have the ability to move point to point and lift.
Now suppose we have an agent who has no way of moving point to point. By acquiring a wheelchair, this person obtains a way of moving point to point. It is important to do this because moving point to point is an accomplishment that interfaces with many other accomplishments and more broadly with many other methods, but not with that of moving between floors, at least not if all that is provided at a location are stairs. In order for moving point to point to be a way of moving between floors something like a ramp is necessary--something that obviates the need to lift. With a ramp, moving point to point is an accomplishment that interfaces with the desired accomplishment; it became of way of moving between floors and thus part of a method for, e.g., hand delivering memos to one's boss.
Now let us consider the relation between walking, walking with crutches and using a wheelchair. Here we include in walking the variation that results in climbing stairs. An agent equipped with crutches is given a new way of walking. That is, walking is not itself an execution, but an accomplishment that standardly involves the execution of a number of movements in a coordinated way. Someone who can execute some but not all of these, or lacks strength or balance, can accomplish the same thing using crutches. Walking is an interfacing accomplishment for many methods and in a very wide range of circumstances. For most of these, walking is first and foremost a way of moving from point to point. Thus, an agent equipped with crutches will interface the various walking-involving methods at the same point as the individual who walks without crutches.
In contrast, the wheelchair user is not provided with a new method of walking, but a new method of moving from one point to another that does not involve walking. Using a wheelchair does not interface with the various walking-involving methods at the same point as walking. It interfaces rather at a slightly more abstract level, precisely that of moving point to point. This is more abstract because both walking and wheeling are ways of moving point to point.
Consider a task such as delivering inter-office memos. Suppose the standard method of doing this in a company is hand-delivery; that is, the messenger picks up the memo from the writer, walks to the office of the intended recipient, and hands it to her. The circumstances at the locations determined by the paths to be taken need only enable walking. This method is essentially unchanged if a crutch-user user becomes the messenger. In the case of the wheelchair user, the paths must enable point-to-point locomotion by wheelchair; typically at least, a wheelchair-accessible path is one that enables walking. Designing and building walking paths in such a way that also renders them wheelchair accessible--and this may, of course, include providing ramps--enables a larger, more abstractly characterized, range of interfacing accomplishments. Providing crutches, while a good thing, does not in this way enable a large range of interfacing accomplishments at a location. Rather, it changes the circumstances of an agent, leaving wider circumstances as they were. Notice that is even more so with respect to providing prostheses, such as artificial legs.
The principle that urges such design we call generic interfacing:
Let us look at another case, one closer to the central focus of the Archimedes Project, a research project at Stanford University whose mission, in part, is to help people with disabilities to communicate and to have access to information through the development of computer technology (see box).
Consider a computer. The standard method for inputting data to a computer is by typing on a keyboard and moving a mouse that are properly attached to the computer. We have:
moving fingers 
typing inputting data
creating/altering files (many
applications)
Keyboards were designed as they were to enable typing as the standard interface accomplishment for a wide range of computational accomplishments and methods. But by using a head-stick, holding a pencil with one's teeth, using one's feet, etc., one can fit into this method at the typing node--that is, all these are modes of depressing the keys in such a way as to bring about all the required computational events. Use of speech recognition technology, on the other hand, allows one to interface this method at the inputing data node, bypassing typing altogether. It renders talking an interface accomplishment relative to the same extremely broad class of accomplishments and methods for which typing and mouse moving were the sole interfacing accomplishments. Thus, the provision of speech recognition technology provides for a wider class of users by enabling a larger class of interfacing accomplishments.
Finally, let us return to the case of the epistemic accomplishment of obtaining information from a display on a monitor. We were supposing our blind reader was dealing with a not quite modern system, in particular a pre-GUI system, one in which all the information to be displayed could be (and was) displayed on the screen via text. In the world after the GUI revolution, provision of screen readers and/or Braille transcribers is no longer enough to render hearing and feeling adequate interface accomplishments, relative to the new, even wider range of accomplishments and methods available to the sighted.
Interaction with computer systems is aimed, first and foremost, at interfacing with the meaning of or information carried by files--which we use as a maximally generic term for data structures. The principle of generic interfacing suggests the following design principles with respect to computer systems: make both input and output as device-neutral as possible, that is, bypass as much as possible the requirements (on executions) of particular peripheral devices. This is typically accomplished by providing alternative peripheral devices enabling other input/output modes. But this depends, in turn, on the form in which the information is carried being accessible to those modes. To the extent that the GUI revolution narrows the interfacing accomplishments for interacting with computers to visually picking up graphical information and pointing with a mouse, it represents a violation of the principle of generic interfacing.