Towers, Grids, Depth and Distance: Spatial Metaphor in Deck Manuals, Interface Specifications and Operator Testimony, 2031–2046

1. Introduction

An operator who jacks in does not look at a table of data. The deck renders what the operator's trodes carry as an environment, and the operator works in it by moving, approaching and entering. Operators talk about the work in the same way. A system is a tower with floors, a database lies deep, a node is near or far, and a route runs across a grid. The words are so ordinary in the trade that operators seldom remark on them. Conventions of depth and scale in matrix visualisation have been catalogued (Salaam-Novak, 2041), and the history of decks and their manuals has been told (Kilbride, 2038), but the metaphors have not been read for what they ask operators to assume.

The matrix is often described as a shared hallucination, an experience that many operators have at once and agree on in its main features, produced by the deck's rendering of data that has no spatial form of its own. The spatial form is therefore supplied by the deck and by the operator's reading of it. The question of this essay follows from that fact. If the environment is an imposed form, then the metaphors that govern it are choices, made by designers and adopted by operators, and they may help or mislead. We ask what they do. A philosophical sketch has treated the matrix as a place (Lévesque-Arno, 2042); we take the more modest view that it is a rendering that is read as a place.

Our thesis has three parts. First, deck documents and operators use different families of spatial metaphor, the documents favouring the grid and the operators favouring depth and distance. Second, each family carries expectations about data that the display does not guarantee. Third, we put forward as a hypothesis, which our corpus can only suggest, that the difference between document and operator is itself a source of error, since an operator trained on a manual's grid may go on to behave in the environment as a traveller. We write in 2047, and the corpus runs from 2031 to 2046. Our argument is interpretive. It draws on counts of the corpus but does not rest on them.

2. Corpus and Approach

The corpus has three parts. The first is 23 deck manuals, supplied by Ono-Sendai and by four other makers whose names we withhold at their request, covering the period 2031 to 2044 and including successive editions of the manual for the Ono-Sendai Cyberspace 7 (Ono-Sendai Deck Engineering Research Group, 2044). The second is 14 interface specifications, internal documents that define how a deck addresses and renders locations in the matrix and that Ono-Sendai released to us for this study (Ono-Sendai Deck Engineering Research Group, 2043). The third is 38 transcribed accounts from 31 operators, gathered between 2041 and 2046 in interviews and in published accounts by the operators themselves. In all, the corpus has 75 documents and about 602,400 words.

Ono-Sendai's role needs stating. All 14 interface specifications and the Cyberspace 7 manual editions come from Ono-Sendai, which chose what to release. One author, Oduro-Blake, is employed by the Ono-Sendai Deck Engineering Research Group, which supplied them. He advised on reading the specifications and did not code documents, and the other authors have no tie to the maker. Readers should treat the account of how decks render space as the maker's own.

This corpus is not a sample of everything said about the matrix. It contains the documents that makers were willing to release and the operators who were willing to be interviewed, and it is biased toward the Sprawl and Chiba and toward operators who had survived their careers. Early manuals are scarce, because the first decks were documented sparsely and much was lost (Belhadj, 2044). We describe these limits again in Section 6, and we ask readers to keep them in mind throughout.

We read the corpus in two passes. In the first we coded each document for the presence of four families of spatial metaphor: vertical (towers, levels, climbing), grid and plane, depth (descent and layers) and distance and travel. A document was counted for a family if it used the family's vocabulary at least once to describe the matrix or its contents, and not simply the deck's controls. Table 1 reports the counts. Presence counts are sensitive to document length. The manuals are the longest documents, averaging about 13,843 words against about 4,928 for operator accounts, so length alone would raise the manual counts, and the differences we report between manuals and operators run in the opposite direction for depth and distance. Many documents use several families, and 24 of the 75 use all four. In the second pass we read closely for what each metaphor was used to say, and the argument rests on that second reading. We paraphrase testimony and identify operators by code. We make no claim about the structure of the matrix itself, and we paraphrase the commonplace descriptions of the matrix as a hallucination without relying on any particular wording.

3. The Grid in the Manual

Manuals and specifications use the grid more often than operators do. It appears in 91% of manuals and 93% of specifications, against 47% of operator accounts. The preference follows from the deck's design. A deck addresses locations by coordinates, and the interface specifications define a lattice of cells in which every renderable object has a position (Ono-Sendai Deck Engineering Research Group, 2043; Wakabayashi-Cole, 2039). A manual that must teach an operator to specify a location has little option but to describe a grid.

The manual's grid carries a lesson beyond addressing. Several manuals explain that objects close together in the lattice are related, that clusters of cells belong to the same system and that an unoccupied region is empty. Readers learn to treat adjacency as meaningful. The specifications are less committal. They say that placement in the lattice is determined by the deck's allocation of rendering resources and by the order in which objects were loaded, and that adjacency has no necessary relation to the objects' function (Crewe, 2040). The corpus does not show how many operators read the specification, and the accounts below suggest that some rely on the manual alone.

We therefore read the grid as an aid with a hidden cost. It makes the matrix teachable and addressable, and it encourages a belief that the layout reflects the organisation of the data. Several operators in our corpus reported moving to a location because it lay next to one they knew, and finding that it belonged to an unrelated system (T-04, T-17, T-22). These accounts are anecdotes. They are seven of the 38 operator accounts, cited in this essay as illustrations of what the readings describe (T-04, T-17 and T-22 on adjacency, T-09 and T-31 on depth, T-12 and T-26 on distance). We did not code the 38 accounts for mistaken inferences, and we cannot say how many contain none.

4. Towers and Depth in the Operator's Account

Operators use depth more often. It appears in 87% of operator accounts and 52% of manuals, and the vertical family appears in 76% of operator accounts and 74% of manuals, about as often in both. Operators describe a protected system as a tower, speak of climbing it and of reaching its upper floors, and speak of valuable data as lying deep. The two images work together: height marks the hierarchy of a system and depth marks how guarded its contents are. In the accounts we read, operators often move between them without marking the difference.

These images carry judgements of value and difficulty. A high floor is more important than a low one, a deeper layer is better defended than a shallow one, and effort is expected to rise with the climb. Operators acted on these expectations. Accounts of intrusion in our corpus describe operators who gave up on a shallow-looking target because it seemed unlikely to be worth the effort, and who spent extra effort on a deep one because depth promised value (T-09, T-31). Neither inference follows from the structure of the target. Both follow from the metaphor.

A study of operators' perceptions of intrusion countermeasures, based on interviews with operators (see the study of ICE topology in the references), reports that operators read the rendered form of ICE as a sign of how the code behind it is built. If so, then for ICE the display gives operators something that they can legitimately read, and the metaphor of depth works where the display has been built to support it. We do not use that study for any more than this. Our point is limited to locations and data, where the corpus gives no reason to think that the display provides comparable support.

5. Distance and the Cost of Travel

Distance is almost as common as depth in operator accounts, and it is the family most tied to the manuals' range parameters. It appears in 82% of accounts and 65% of manuals. Operators speak of a node as near or far, of long routes and short ones, and of having to travel to reach a distant system. The language is frequent in manuals as well, where it appears in connection with range parameters that limit how far from a starting point an operator's actions may reach.

In the matrix, distance is not a metric. The specification defines it as a count of rendering steps between two locations, a quantity that depends on the deck's routing and not on any geometry (Sirvent, 2042). Operators nonetheless treat it as a measure of effort, risk and time, and the relation they assume is not guaranteed. A node that is a few steps away may be inaccessible, and a node that is many steps away may open at once. One operator who had worked for years described learning to distrust the sense of distance (T-12), and another admitted never having learned to (T-26).

The travel metaphor has one further consequence. It makes the operator an agent moving through a space and not a person issuing requests to a system. This framing makes moving feel like action and waiting feel like loss, and it may encourage operators to prefer visible movement to the passive checking that a careful intrusion requires. We suggest this as a hypothesis, which the corpus supports only weakly, and which we have not tested.

6. Counter-readings

Language as convention. A sceptic may hold that operators use spatial words as trade shorthand, as people say that prices are high or that a mood is low, and that nothing about their thinking depends on them. We cannot refute this from the corpus alone, and shorthand and inference are not exclusive: a conventional phrase can still carry its entailments. We note only that the seven accounts cited above describe actions that followed the metaphor, that we did not count accounts in which no such action occurred, and that the shorthand reading must explain why operators in those seven acted as though adjacency, height and distance meant what they said.

Literal hallucination. If the matrix is experienced as a space, a tower in it is a tower and a distance is a distance, and we are describing description. This objection has force. The experience is spatial, and operators do move. Our reply is that the experience is a rendering of data that has no spatial form, as the specifications say, so the question is what the rendering asks the operator to infer. An operator can reasonably treat a tower as a tower in the display and still be wrong to treat it as a hierarchy of the data behind it. The metaphors are literal about the display and figurative about the data.

Bias in the corpus. It is present. The manuals are those that makers chose to release, the testimony comes from operators who agreed to speak, and the early period is thin. Of the 24 documents that use all four families, 14 are operator accounts, 7 are manuals and 3 are specifications. The counts describe the documents we have, and a different corpus could produce different proportions. We have tried to make the argument turn on the readings and not on the counts.

Design follows habit. It may be that makers adopted a grid and a vertical vocabulary because operators already thought in those terms, so that the metaphors shape the manuals and not the reverse. The corpus is compatible with this, and we accept it as a partial explanation. It cannot show which came first, because the manuals begin in 2031 while the operator testimony begins only in 2041, so any order we saw between them would be an artefact of how the corpus was assembled. The influence may have run both ways, and our argument does not require that it ran only one.

7. Conclusion

The spatial metaphors of the matrix are not neutral. The grid makes a deck teachable and may suggest that layout reflects organisation. Depth and height give operators a way of judging value and defence. Distance gives them a sense of effort, although the specifications define it as a count of rendering steps. Each carries inferences that the display does not guarantee, and operators who trust them could be misled, as the accounts we cite illustrate.

Training could attend to the gap between the manual's grid and the operator's depth and distance. A manual that taught the grid as an addressing convention, and warned that the specifications attach no necessary meaning to adjacency or distance, might give new operators a reason to check what the metaphors invite them to assume. We offer this as a suggestion for testing, not a recommendation, and we recognise that it rests on an interpretive reading of a corpus whose limits we have stated, and on documents supplied by one maker. A larger corpus and a study of operators' errors would test it.