What is the equivalent of a Turing machine in architecture?
And in what sense do cellular automata "picture" a state of affairs?

A Turing Machine in Architecture?

A Turing Machine == a program. Not machine mechanically speaking - hardware - but machine in the abstract sense - software. These 'machines' takes any set of data or inputs (any data set), reads it (if it can) and spits out whatever its supposed to. A program defines what can be read.

What these rules do is basically independent of the circumstances in which they will play out. By circumstances I mean where or when or on what they will be acting on, e.g. the data can be anything as long as it is readable by the rule set. A program has a certain language, and if it encounters 'words' outside of its specific language structure the program will stop functioning. So at the onset, if a program is to be used, the language it uses must be clear of unreadable data. Cellular Automata are a way of picturing such defined arrays of stuff.

Cellular Automata are a way of picturing how rules effect a field of machines over time. One frame represents how the program has reorganized objects' transformable attributes or variables based on their relation to their immediate surroundings. The states of each individual unit in the array have certain possible states, states being the current array of variables.

Objects with series of affectable attributes are placed in a field as an initial condition and all are embedded with a program in order to read and to act. Beforehand the patterns of attributes to be acted on and their corresponding actions are defined. The program begins and if the program see's a certain pattern of variable states within the cell's vision it performs a pre-defined action which reconfigures that pattern.

It is interesting to think how a program "sees". What is it looking at and in what order? Either way it must be systematic.

By running a Cellular Automata, pictures are formed of possible states of the entire system as a whole, while the governing process only happens locally within the space. Certain local states are predicted so as to act upon them, but equilibriums or periodic patterns or chaotic systems can emerge within the whole even though the cells are acting independantly. Cellular Automata seen as a Video, or stills placed in sequence, allows us to see animate behavior. Not necessarily of physical of objects, but the data controlling them.

Is it possible to account for any type of data in a computational Architecture? Programs can be written to accommodate knowable and definable inputs, the most obvious being loads, specifically dead loads, in a structure.

Engineering could be a type of software which reads our often rediculous plans(or more recently parametric models) to spit out revised drawings. The rule-set is based on the need to resist load. A field could be forces in space which must be transfered to a surface (the ground)

Strictly speaking, Structural engineering at this point in time is not a form of computation. It is top down, it does not emerge meaning the rules are applied to a completed form , the rules themselves do not necessarily generate form. A sort of engineering could take place in design if 'rules for maintaining stability' were integrated and let run on a set of building parts.

It's interesting to note that two different data sets can only shoot out the same answer if the program(machine) is different. take the data set [1,1] the program 'add()' spits out two. The same program on a different data set, say [2,1], gives three. but you can get the same answer if you change the program to 'multiply()'.