From Drawings to Systems
Traditional architectural drawing is fundamentally additive. Designers add marks, erase marks, and overlay independent lines to build up complexity. Relationships exist, but they remain implicit maintained mentally by the designer rather than embedded in the medium itself.
As Robert Woodbury explains in Elements of Parametric Design, the archetypal design medium is pencil, eraser, and paper. The pencil adds. The eraser subtracts. Even when precision tools are introduced, the logic remains unchanged: designers accumulate independent elements and personally manage their consistency.
When CAD software emerged, the medium changed, but the underlying logic did not. Lines became digital, layers replaced tracing paper, and drafting became faster. However, the computer largely remained a representational device. This phase, often described as computerization, digitised preconceived ideas rather than enabling design reasoning.
Kostas Terzidis, in Algorithmic Architecture, draws a clear distinction. Computerization is the act of entering, processing, and storing information. It is extremely powerful for documentation and repetition, but it does not reason, deduce, or explore.
Computerization works effectively for deterministic problems such as drafting, documentation, visualisation, and modification. However, contemporary architectural challenges increasingly involve uncertainty, interdependence, performance feedback, and emergent behaviour, conditions that cannot be resolved through representation alone.
Computerization vs Computation
Computation is not simply the use of computers. It is the use of logical and procedural systems to derive outcomes from relationships and rules.
Terzidis defines computation as determining something through mathematical or logical methods. More importantly, he emphasises that computation is exploratory. It addresses ill-defined problems and extends human reasoning by enabling systems to evaluate, deduce, and generate outcomes.
Achim Menges similarly describes computation as increasing the specificity of information from abstraction. Rather than manually resolving every consequence, computational systems deduce results from initial values and relationships.
This shift is not merely technical. It is methodological. Design moves from drawing representations to constructing systems capable of reasoning.
Parametric, algorithmic, and generative design all emerge from this transformation and represent different ways of approaching architecture computationally.
What Is Parametric Design — At Its Core?
Parametric design begins when an architect stops designing isolated objects and starts designing relationships.
At its core, parametric design is not about visual complexity. It is about explicitly defining how elements depend on one another. Walls, floors, facades, structural members, and services become parts of a connected system.
In traditional drawing, designers manage consistency mentally. In parametric systems, consistency is embedded. Relationships are defined in advance, allowing coordinated change.
The designer moves from controlling individual elements to constructing frameworks of dependency.
As Robert Woodbury explains in Elements of Parametric Design, traditional design operates through adding and erasing independent elements. Parametric design introduces a shift: designers add, erase, relate, and repair. When one part changes, dependent parts are reorganised through relationships rather than manually redrawn.
Parameters and Constraints
Every parametric system rests on two foundations. Parameters define what can vary, including dimensions, quantities, densities, and performance values. Constraints define how elements relate, including alignment, proportionality, containment, and continuity.
Together, they transform drawings into coordinated systems. When one value changes, the system reorganises itself according to predefined rules.
This is why parametric design is often described as associative geometry. Branko Kolarevic describes parametric design in Architecture in the Digital Age as associative geometry, geometry whose behaviour is defined by relationships that can be revisited and revised throughout the design process. Geometry becomes responsive rather than static.
Typical Architectural Use Cases
Parametric design becomes essential wherever coordination is unavoidable.
In Facades
Facades operate as dependency networks. Panels relate to grids. Openings relate to daylight. Geometry relates to fabrication. Density relates to performance.
Contemporary facade systems must respond to environmental data, material constraints, structural logic, and construction workflows. Parametric systems allow these requirements to remain coherent.
Here, parametric design functions as organisational infrastructure.
In Towers
Tall buildings intensify interdependence. Floor plates depend on cores. Facades depend on stacking logic. Structural systems depend on spans. Zoning depends on regulatory envelopes.
Manual coordination becomes unreliable at this scale. Parametric systems allow global changes to propagate systematically.
In high-rise projects, parametric design becomes a mechanism for managing complexity.
To get more insights on the application of computational design in architecture, read our detailed blog.
What Makes Algorithmic Design Fundamentally Different?
A parametric model becomes algorithmic when geometry is no longer the starting point and logic becomes primary.
Algorithmic design begins with a procedure. It focuses on defining step-by-step reasoning that structures how decisions are made. Instead of shaping outcomes directly, the designer organises processes that generate outcomes.
An algorithm is a finite sequence of instructions that performs a task or solves a problem. Arturo Tedeschi defines algorithms as finite, well-defined instructions that return a solution or perform a task, emphasising that algorithms are not inherently digital but reflect structured human reasoning broken into executable steps. These instructions may exist in code, diagrams, or written logic.
Logic-First vs Geometry-First Thinking
Parametric design often starts with geometry and makes it responsive. Algorithmic design starts with logic and allows geometry to emerge. This represents a shift from composing forms to designing decision systems. Buildings become the result of encoded reasoning.
Conditional Logic, Loops, and Rule-Based Systems
Algorithmic design introduces conditional statements, iterative loops, hierarchies, and evaluation structures.
These mechanisms allow systems to filter options, sequence operations, resolve conflicts, and prioritise outcomes. Instead of reacting passively, systems actively reason through scenarios.
As Kostas Terzidis clarifies in Algorithmic Architecture, computation differs from mere computerisation. Computerisation processes and stores information; computation determines outcomes through logical and mathematical methods. It reasons, evaluates, and deduces.
Deterministic Outcomes
Most architectural algorithms are deterministic. Given the same inputs and rules, they produce the same outputs. Terzidis notes that algorithms formalise design reasoning into repeatable logical structures; they do not replace the designer, but operationalise structured thinking. They formalise design thinking and make it scalable.
Examples in Practice
Algorithmic thinking appears wherever architecture becomes procedural. In space planning, algorithms structure adjacency and circulation. In code-compliant layouts, they formalise regulatory and safety requirements. In optimisation workflows, they enable systematic testing and refinement. In these contexts, form is computed rather than composed.
What Exactly Is Generative Design?
Generative design is a strategic use of computational systems to explore multiple solutions. Generative systems produce many outcomes from the same goals and constraints. Designers define objectives, limitations, and evaluation criteria. The system searches the solution space.
Generative design becomes possible because computation is exploratory. Terzidis emphasises that computational systems can address indeterminate problems by evaluating and deducing outcomes rather than simply representing predefined ideas.
In practice, generative design appears in two dominant modes. The first is optioneering. Systems generate diverse configurations for early-stage exploration.
The second is optimization. Systems iteratively refine solutions toward performance targets. Achim Menges describes computation as increasing the specificity of information from abstraction, allowing results to be deduced from initial parameters rather than manually resolved step by step. Generative systems extend this principle into multi-solution exploration.
Despite its visibility in software marketing, generative design is less common in daily practice. It is usually confined to research teams, competitions, and specialised technical groups.
Most firms embed generative moments within broader parametric and algorithmic workflows. Without stable relationships and clear logic, generative systems produce variation without direction.
How These Approaches Combine in Real Projects
In real projects, parametric, algorithmic, and generative design function as layers. Algorithmic structures define decision logic. Parametric systems embody those decisions in coordinated geometry. Generative strategies allow exploration and refinement.
A project may use algorithms for zoning, parametric systems for building components, and generative solvers for performance testing.
What is often labelled “generative design” is typically a mature parametric and algorithmic framework with targeted exploration. Generative methods amplify structured thinking.
Which One Should Architects Learn First — And Why?
For architects with limited time, the learning sequence matters.Parametric thinking should come first. It builds sensitivity to relationships and coordinated change.Algorithmic thinking should follow. It develops procedural reasoning and abstraction.
Generative thinking belongs later. It functions effectively only when systems are stable.
1. Skill Maturity vs Project Complexity
As project complexity increases, architectural work shifts from form-making to system management. Larger teams and higher performance demands require stronger computational frameworks. Computational maturity must grow alongside responsibility.
2. Common Beginner Mistakes
Common mistakes include jumping directly into generative tools, confusing variation with intelligence, building unstable parametric models, and optimising poorly defined problems.
These practices produce visual complexity without structural clarity. Computational design is about constructing better systems.
How To Get Started?
Most architectural education still prioritises additive workflows. Far fewer programmes train designers to build systems, formalise relationships, and encode logic.
Yet contemporary practice increasingly demands coordinated change management, performance-driven workflows, and data-informed decision structures.
Developing computational literacy requires progression from parametric foundations to algorithmic reasoning and finally to generative exploration. Programmes such as Novatr’s Master Computational Design Course are structured around this progression and grounded in architectural application.
TABLE OF CONTENTS
