A recurring theme of this newsletter is that most new ideas and innovations in the construction industry have been tried, often much earlier than you might expect.
I concur completely with Jim's comment on ResCheck and Comcheck. Narrowly defined, 2-dimensional software programs that deal with one aspect of building performance are also open to "gaming the system" with bad inputs. The human element in the building inspection process for energy conservation requires inspectors who are trained and motivated to be picky about insulation installation, air-sealing, etc...
On a related note, the Architectural Registration Exam graphic software is fully automated with respect to evaluating a candidate's "design" for code compliant building elements like egress, stair proportions, furniture placement. I believe the code for this dates to the 1980's. It's crude, but I'd argue that it's sufficient for drilling my profession into thinking about objective results; i.e. making sure a building has minimum life-safety parameters.
Scaling up ARE-like software to a BIM model would require standardization of software inputs and outputs. I suppose machine learning could be applied to the evaluation of floor plans in .pdf format. Where's the incentive in that?
Many thanks, very interesting reading. The challenges highlighted could be applied to many other areas where digitization is tried, but turns out to be harder than it seems.
There is some activity from Google on this. For their own internal use, they created a Digital Buildings model language. More importantly, an Alphabet spinoff, Flux, tried to solve these issues in 2017–2020, before throwing in the towel, as noted here: https://parametricmonkey.com/2020/06/03/rediscovering-flux/ . As a story, it confirms the points you make in your article.
I concur completely with Jim's comment on ResCheck and Comcheck. Narrowly defined, 2-dimensional software programs that deal with one aspect of building performance are also open to "gaming the system" with bad inputs. The human element in the building inspection process for energy conservation requires inspectors who are trained and motivated to be picky about insulation installation, air-sealing, etc...
On a related note, the Architectural Registration Exam graphic software is fully automated with respect to evaluating a candidate's "design" for code compliant building elements like egress, stair proportions, furniture placement. I believe the code for this dates to the 1980's. It's crude, but I'd argue that it's sufficient for drilling my profession into thinking about objective results; i.e. making sure a building has minimum life-safety parameters.
Scaling up ARE-like software to a BIM model would require standardization of software inputs and outputs. I suppose machine learning could be applied to the evaluation of floor plans in .pdf format. Where's the incentive in that?
Thanks Brian, wonderfully written. Keep posting great content. Best Wishes from Australia.
Many thanks, very interesting reading. The challenges highlighted could be applied to many other areas where digitization is tried, but turns out to be harder than it seems.
There is some activity from Google on this. For their own internal use, they created a Digital Buildings model language. More importantly, an Alphabet spinoff, Flux, tried to solve these issues in 2017–2020, before throwing in the towel, as noted here: https://parametricmonkey.com/2020/06/03/rediscovering-flux/ . As a story, it confirms the points you make in your article.
This is an interesting point I hadn't considered, thanks.