This is a self-translated/enhanced summary of the presentation made on September the 27th. Feel free to leave constructive criticism!
This doctoral research proposes the modelling of the creative processes of architectural design through high abstraction level systems based on informational thinking strategies. The models are virtualized into interactive digital constructs developed with visual scripting tools for the integration of all systems of the architecture process.
The origin of this research comes from the merge of an interest and an intuition:
We are interested in architecture as a thinking system whose main goal is the creation of human space.
We believe that abstract thinking in architectural design can benefit from informational thinking.
To achieve the merging of this intuition and interest, the proposal works by embodying space as an abstract object into contemporary digital tools by means of informational thinking strategies.
This approach is different from current dynamics in most of the research lines and work fields on digital tools in that:
These are specially focused on the technical, management, form-generation and experimental fabrication aspects.
The computational work dynamics are generally based on the use of digital tools on the work object. This bottom-up approach produces a strong instrumental dependency we strive to dismantle.
Architecture as a system
Besides emphasizing our interest in abstract thinking about space, we have to specify further in that the architecture creation system object of our research is the one that adheres to the following sequence:
Spatial thinking (self referential goals)
Document production (production instructions)
Communication with the production agents (translation of the instructions into material actions)
Material building (realization process)
With the particularity that as this sequence unfurls, iteration loops happen.
We detect several points in the mentioned system that we consider problematic:
Abstract spatial thinking is usually produced on analog supports and has a tremendous level of abstraction (sketches, work models, diagrams, etc.)
Tech embodiment introduces a kind of complexity that is more and more difficult to cope with and sync with abstract thinking on space.
The production of graphic documents (2D and 3D) has been informationalized through a system of tools, processes and methodologies that go from computational design to BIM.
In our opinion, these issues are cause and reason of the distance, disaffection and wastefulness between the two extremes of the architectural process we are working on.
From the transdisciplinar position of our research we propose the base hypothesis:
Like programming languages, that embody the thinking, design and production processes in software engineering; architecture processes can also be classified in high abstraction level and low abstraction level systems and both types can be integrated with an intermediate compilation system.
Compilation from HL to LL
In software, high level systems are programming languages. They are suited for humans and their components have a high level of abstraction; whereas low level languages take care of the communication with the machine but are highly concrete and of very difficult intelligibility for humans.
Between high level and low level systems the compilation or assembly processes take place. These are produced by an intermediate system that traduces the instructions given in the high level system into the low level system so that the machine can execute them.
Low level to high level architecture
In architecture, low level systems are the building and technical documentation objects that are issued for construction companies. They embody material concreteness, they give the precise and unequivocal instructions to the production agent.
High level systems are those that allow to organize thinking in a strongly abstract way previous to material concretion. Be it about spatial interests, programmatic requirements, heritage implications, social conditions, poetic meaning… any and all aspects that could be part of architectural thought but do not need to come from a material approach.
Compilation of High abstraction level analog systems
Currently high level architectural systems are supported on analog media or in best cases are digitized or computerized, but their informational potential is still to be used.
The analog-digital problem
The current compilation process between high level and low level in architecture is producing two very distinct phenomena which can barely be adressed as problems:
On one side architecture is being produced with a straightforward use of low level systems. In our opinion, the absence of high level systems in these processes makes the result loose the complexity that is derived from abstract thinking.
On the other side, when these high level processes indeed take place, it is through/with analog media. The change of level is done by human compilers that can be either the agents in charge of the initial high level processes or other agents that have already been denominated superusers  . In any of these two cases, the creative process has one or several points of rupture.
From the understanding of the duality of architectural thought systems, and looking forward to transcend the conceptualization of “language” towards which the parallelism with software systems would lead us, we would enter the realm of programming paradigms.
It is not operative to follow the analogy and choose a programming paradigm, not even the use of visual scripting tools for the prototypes and applications development is enough justification. It should suffice to say that we will discuss how the modelling of architectural thought processes we propose has strong similarities with the functional and with the object oriented programming paradigms.
Models and strategies
Our research is developed in two parallel blocks:
First the design principles. An open catalogue of informational thinking strategies that easy the construction of the high level of abstraction objectual models.
Second the proposal of a formalization of these high level models, which we will call henceforth paratypes, and their prototyping and testing with digital visual scripting tools.
Paratypes are cognitional models based on the following premises:
Their goal is modelling high level of abstraction architectural thought processes.
They enable the hybridization of digital and analog objects/constructs.
They have an informational condition. Thanks to it they can be compiled into low level systems and produce then an operative integration of the complete architectural process.
We propose the modelling of the constructs as DAGs, directed acyclic graphs.
They respond to a process dynamic.
They take advantage of the dataflow graph’s logic.
They enable the understanding of the relations among the network of differential objects.
Though cycles might be produced, these are not automatic and the graph has always a determined state within its metastability.
The modelling system has two basic components: objects and operations.
The models make explicit the relations among the components and the operations they intervene in or are subject to.
The graph entity expresses the structure of the process with its linearity and iteration.
The variability of the objects and the operations among them makes the resulting model-graph a metastable entity.
System components are objects and operations. They can be of three different kinds:
Computational, which we will consider always internal to the machine and indifferent from their support.
Human, which we consider always internal to the human agent and without the need of an external physical support.
Analog, which in having a physical support, which itself should be inert to digitization or computerization operations, is also indifferent to their human or digital condition.
Objects have complexity levels that correlate to the Noise-Data-Information-Knowledge-Wisdom pyramid . We take the three basic elements from this hierarchic structure, data, information and knowledge to articulate the complexity levels of the model components. Though the three concepts have been and still are subject to discussion in the fields of information and knowledge sciences, and will be treated in depth in the dissertation, we can adhere to the following definitions:
Data: Is structured content but without relation to any other external object.
Information: Is data structures related among themselves to produce a higher order object.
Knowledge: Is information structures related among themselves to produce emergent information and/or knowledge when cognitional operations are done upon them.
We have classified the operations in two generic types:
Analysis: Produce new partial, and generally lower level of abstraction, informational entities from the original operands.
Synthesis: Produce new informational entities of an equal or higher abstraction level than that of the original operands.
The breach existing between high abstraction level systems, which allow us to instrumentalize abstract system, and digital low level systems is an operative abyss. Digital tools used till now, focused in low level objects, be them geometric or constructive, does not facilitate abstract spatial thinking and even less modelling it.
Strategies allow us to understand that working on high abstraction level systems we dwell in the realm of virtuality, of the exploration of possibilities, control and interactivity. Through the mechanisms of interaction and control is how the compilation of low level systems, the documentation for the material production, is based on the knowledge embedded in the paratypes. The compilation process is then reduced to the actualization of the virtuality encapsulated in the paratypes.
To develop this modelling processes, the articulation of the thought structures and the work on them once objectualized, it is necessary to deploy thinking strategies that should enable us to build for ourselves an operative frame of reference.
The work-in-progress catalogue of strategies is as follows.
Without the thought strategies, the resulting models would be mere general formalizations to guide parametric design processes aimed towards lower level systems.
With the strategies, we see that in the model components it is possible to frame the higher level systems as well as the lower level constructs, enabling the construction of the paratypes as virtual containers of the architectural design as integral process.
Advantages of the paratypes
The main advantages of the paratypes are the increase of control over the design intellective processes, the construction of design spaces  and the emergence of the design as an integrated process.
In the current state of our research, we can postulate with some conviction that the advantages of producing interactive models are:
- Increase of control over the design as a complex system. The thought process whos goal is the production of a document-object changes into an integrated system in which the control points we establish allow its ad-hoc configuration.
To conclude we can understand paratypes as constructs that model and instrumentalize the thought processes through the operation and control over their constituent information.
Deutsch, Randy. 2019. Superusers – Design Technology Specialists and the Future of Practice. New York: Routledge. ↩
Zins, Chaim. 2007. “Conceptual Approaches for Defining Data, Information, and Knowledge.” Journal of the American Society for Information Science and Technology 58 (4): 479–93. https://doi.org/10.1002/asi.20508. ↩
space as in vector space not as in the space of the atrium. ↩