Dreamcatcher: The End of Design As We Know It?
Noted designer Dan Saffer wrote a good piece about the future of design and what changes could be coming called The End of Design As We Know It. Dan says:
"These changes will be drastic and design will never be the same afterwards. The canary in the coal mine is Autodesk’s Project Dreamcatcher...To many people this is a bleak, grim, oh-shit-there-goes-my-job, future."
It's nice to being working on a project that captures attention and generates strong discussion. If you haven't heard of Project Dreamcatcher before, the high-level pitch is that you can supply the computer with a bunch of design goals and have the computer develop a bunch of options that meet your criteria for further refinement.
Dan goes on to make some good points on how this could work to benefit the designer as well as things that the designer will still be good at. A while back Erin, Francesco and Mark from the Dreamcatcher team did some research on how designers might interact with such a system and documented it in a paper called Parameters Tell the Design Story: Ideation and Abstraction in Design Optimization. The main finding from this research is summed up as:
"We found that the computed optimum was often used as the starting point for design exploration, not the end product."
In other words, designers would use such a system to generate and explore solutions, possibly finding unexpected solutions. Some call this design optioneering. In this sense we can redefine the meaning of CAD. With typical workflows, most of the design is still happening in the designer's head and the computer is helping to document those ideas - CAD could mean Computer Aided Documentation. Now with the computer actually doing design work, we can truly realize Computer Aided Design.
In common workflows, every design that must be delivered to a client takes time, effort and money so clients usually only get a couple of options.
How does Design Optioneering Work?
Let's look at building design. There are a number of constraints to be considered, including:
- site utilization
- structural design
- building form
This many requirements can be very complex to design for and creates a huge number of possible outcomes. So huge actually that we can't even imagine a small fraction of the possibilities due to cognitive limitations. Using high performance computing and big data analysis techniques, many more design alternatives can be explored for a problem space.
To show how this might work, we can look at an architectural project researchers from the University of Southern California (USC) did at the Autodesk IdeaStudio with similar constraints. They developed a model and automated the process of exploring the design space. In their work, the USC researchers analysed the options to show how the different results met the criteria with both visual and quantitative results.
We could see a place in the future where designers could easily modify the results for further exploration. Instead of choosing between options A, B and C, designers could further explore the design space between options A and C by changing some design parameters. Of course, designers could modify the resulting CAD file to further customize and finalize the design.
What do designers think?
In looking at how designers would work with Project Dreamcatcher, what kind of feedback did the team get?
“instead of starting with nothing, you start with something…your optimum gives you a starting hunch.”
“slight variations to form, to a designer’s eye, are either elegant or fat.”
“Good design has inspiration to it...if you have that vision you can encode it and parameterize it and explore it further. Now we have a rich flora of options.”
This indicates that such tools are beneficial and could help to improve the design process. Sounds great but what are the challenges?
Dan pointed out that these kinds of workflows would require some different thinking:
"In order to provide the algorithm the right information to suggest solutions, you have to be certain you’re solving the right problem."
Working differently through Design Optimization
One of the ways the Dreamcatcher team thinks about this is with design optimization. Design optimization puts the emphasis on defining the design problem. Design optimization is important to sustainability. Users of design optimization are making buildings and products more structurally sound with less building materials.
In a typical CAD workflow, one follows these steps:
- select or redesign
With Design Optimization, the steps are:
- define the problem
- generate and explore
- select or redefine
One participant in the study compared these two processes:
“The typical design workflow is to design then throw to the analyst. Redesign. And then keep playing catch. It’s inefficient. [Design optimization] captures the criteria that are important to you then [you] have the cloud process all the permutations.”
Thinking differently will not be without challenges. This could require new skills and language. One study participant said:
“1200 variations, you’re not being an architect any more. You are a computer programmer."
Another stressed that the controls need to be easy to understand, control and help you produce good results:
“parameters need to tell the design story.”
This is great feedback and guidance. This is one of the main reasons we do research. Autodesk Research and the Dreamcatcher team are up for the challenge. As Dan said in his blog post:
"Lastly, the way all progress moves forward is by someone thinking differently, trying something that no one else has tried, breaking the established traditions and rules. This is the same in design as in any other field."
Project Dreamcatcher could be the end of design as we know it and like when we transitioned from paper documentation to computer documentation of designs, we may wonder how we ever got anything done before.
You can follow this conversation by subscribing to the comment feed for this post.