What Computational Design Is

Here is the whole shift in one image. To make a row of columns in CAD, you draw a column and copy it twenty times; if the building gets longer, you draw more, by hand. In computational design you instead say: place a column every 4 metres along this curve. That sentence is a rule. Now the curve can bend, the building can stretch, the spacing can change - and the columns re-solve themselves, correctly, every time.

You have stopped designing the result and started designing the process that produces the result. This is why it's called computational: you are authoring a small computation - a set of instructions and relationships - and the specific geometry is just its output for the current inputs. The geometry becomes a consequence of your logic rather than a thing you place. Almost everything else in this course is a richer version of this single move.

These three terms get used loosely, so let's pin them down. Parametric means the design is driven by parameters - named values (a height, an angle, a count) - with relationships wired between them, so changing one ripples through the rest. That's the everyday heart of this course: Grasshopper is a parametric modeller.

Computational is the broader umbrella: designing by defining any logic or algorithm - parametric relationships, yes, but also loops, conditionals, and simulation. Generative narrows in the other direction: systems that generate many design options from rules, often steered by goals or evolution (you'll meet this in the optimization module). In short: all parametric design is computational; generative design is computational design pointed at producing and selecting options. Don't over-police the vocabulary - but know that when someone says 'parametric', they usually mean parameters with live relationships, and that is exactly where we begin.

In Grasshopper - the tool at the centre of this course - you don't build geometry by hand. You build a definition: a visual graph of components wired together, where data flows left to right from inputs (sliders, curves, numbers) through operations (move, divide, extrude) to output geometry. That graph is your design. The 3D shape you see in Rhino is merely what the graph currently evaluates to.

This has three consequences that make the effort worth it. First, flexibility: the design is editable at the level of intent - nudge a slider, get a new-but-consistent result. Second, reuse: a good definition is a reusable engine - the same panelization graph works on a different facade tomorrow. Third, exploration at scale: because the design is parametric, you can sweep the parameters and generate dozens or thousands of variants automatically. You are no longer making one design; you are making a system that makes designs.

Computational design is one of the most in-demand technical skills in architecture, and the gap between demand and supply is real. Complex facades, large-scale panelization, BIM automation, environmental optimization, digital fabrication - all of it runs on these methods, and practices from Zaha Hadid to local firms hire specifically for Grasshopper fluency. A student who can show a parametric project stands out immediately; there are whole roles ('computational designer', 'design technologist') built around it.

But it rewards thinking, not memorising components. The value isn't knowing which button divides a curve - it's being able to look at a design problem and see the underlying logic: what are the parameters, what are the relationships, where's the rule? That is a learnable habit of mind, and building it - deliberately, lesson by lesson - is exactly what this course is for.