Bernstein on the future of architecture

Dec 16, 2010
  • Article by Maitiú Ward
  • Designer

Yale lecturer Phil Bernstein leads industry strategy and relations for Autodesk’s architecture, engineering and construction division, where he is a vice president. Responsible for the future vision of the company, he has some strong views about where design and construction in the built environment is heading. Architects, in his view, will either embrace that future, or simply get left behind. AR editor Maitiú Ward spoke to him at Autodesk’s 2010 conference in Las Vegas.

Maitiú Ward: We’ve seen a lot of uptake of BIM in Australia, but many practices that have embraced it are really struggling with the workplace culture aspects, the workflow aspects.
Phil Bernstein: This is just my opinion, but the industry in Australia is trying to implement these technologies without having the broader discussion about what the process implications are. In the US, the uptake of the technologies occurred simultaneously with a deep examination of the structural challenges of the industry, addressing questions of productivity, limited profitability, problems of project delivery, project organisation, sustainability, opportunities for digital fabrication. In Australia the whole thing started with a question: “Which technology platform am I going to use to create my same old deliverables?” In the US, the discussion has addressed how we can change methodology to increase productivity, how we should change the means of delivery because the technology has made the relationship between design and construction different. So in the States, with adoption rates in some areas pushing 60 percent, we see architects experimenting with different kinds of delivery structures, new insurance models, new contract models, all kinds of new processes.

MW There are all kinds of questions there around ownership – who owns the model, who’s liable, who bears the risk?
PB In the old structures, these are big issues. In the new structures, they’re less so. In a project delivery structure that’s been optimised to really take advantage of a building information model, nobody gives a shit who owns the data. Because the project delivery structure has been remediated to change the nature of the risk proposition. And people are starting to embrace this very counter-intuitive concept that by embracing more of the risk of the project, they reduce the overall risk of the project.

MW Jeff Kowalski talked about predictive modelling for foot traffic flows and that kind of thing – but Paul McRoberts talked about the need for more capability and less complexity. Behavioural modelling is pretty grainy detail that necessitates a particular level of expertise, and that field is probably not familiar to most architects.
PB I agree, so the bad news is exactly what you just said. But the good news is that analysis is very subject to algorithmic description. For example, the media and entertainment division needs simulation algorithms to make people behave naturally in gaming platforms. We can take that same set of algorithmic strategies and translate it into a simulation environment that does fire exiting. That sort of stuff won’t run on your desktop computer, but it will run on our cloud. It’s completely analogous to energy calculations. Energy calculations are very computationally intense and architects don’t know anything about them.

MW Well it’s one thing to create an empirical data set from solar patterns, it’s another thing to create an empirical data set from human behaviour.
PB I was the project manager for the domestic airport in Washington DC, and we had a consultant who had built a set of very simple algorithms using AutoLISP. You could plug in the flight schedule for a day, and he would disperse into his model a bunch of little dots and they would run through the terminal and you could figure out whether, at this particular moment in time, that escalator was big enough to get all those people down to bag claim. And that was using AutoLISP routines. If computational power is essentially unconstrained by the cloud, the simulation problem is trivial.
Our research group at Autodesk recently wired up our office in Toronto with sensors so they can tell when anybody is in a work cube and how much energy they’re using. They’ve created this very complex data collection mechanism that basically builds an energy map of the building relative to the occupancy of the building. So they know that when someone is in cube 623, they’re there for 18 minutes, they use their light for 22 minutes and their computer for 32 minutes. Then we’re aggregating all that data; that’s harder.

MW There’s been a lot of talk in Australia about a common BIM file format, is that something you guys are looking at?
PB There’s just no way there’s going to be a common BIM file format. Autodesk has spent, by my rough calculations, US$600 million on this BIM thing. The theory of building representation that that data format represents is the family jewels. Interoperability? All for it. Clear data exchange formats? All for it. Open BIM data formats? No way. We think our walls work better than ArchiCAD’s walls, and our parametric relationships between walls and floors are better than Graphisoft’s parametric relationships. No way are we letting all that stuff out for the world to see. Politely speaking. I’m sure some of my colleagues would put that in more polite terms…

MW That augmented reality stuff – how serious are you guys about that?
PB Think about some kid who’s 15 years old. Fifteen years from now, he’s going to be the designer, the builder or the client of the future. My generation grew up reading plans, sections and elevations and we’re pretty good at it. These guys grow up flying F18 fighter planes on iPhones – they’re going to expect to interact with designs in some sort of augmented reality format. So, we have an entire division that does nothing but work on that problem – the Media and Entertainment Division. We just steal their stuff. We’re already doing it, if you look at our project like Showcase, which is our realtime walkthrough visualisation platform. Take a Revit model, plop it into this thing and it’s basically a gaming engine. Things unquestionably have to go in that direction. The days of plans, sections and elevations are dead. They’re over.

Nest Collection at Interstudio

Nest Collection, designed by Swedish brand Form US with Love, embodies the concept of giving humanity a chance to take a break.

Conversation • 0 comments

Add to this conversation


16 Dec 10 at 5:21 AM • Adrian

Interesting, coming from the company that has for so long and up until recently, been the lead foot in the industry when it comes to BIM. At the risk of turning this into a Revit vs the rest discussion, Revit is a decent product, but it’s a shame I feel like a robot when using it, compared to ArchiCAD or Vectorworks. It’s true Revit has some advantages over the others, but we still need to produce sharp 2D drawings, and in that respect Revit fails spectacularly, as far as I can tell.

What PB should have said is AutoCAD is dead, but he can’t admit that. Plans, sections and elevations will always play a part in construction documents and will certainly never be dead.

16 Dec 10 at 6:35 AM • Mark

Adrian, check out wikipedia under ‘origins of bim’. Autodesk have not been the lead foot, archicad invented it! Autodesk just called it something else ie BIM instead of virtual building and admittedly have marketed BIM so much better than Graphisoft in recent years but not always.

Anyway…back to the article, typical that Autodesk doesnt want a open format because they want to corner the market just like they did with DWG. We had to live through the years of using ArchiCAD while everyone else used AutoCAD just because it was popular when ArchiCAD was so so much better for architects. Now there is a risk this is going to happen with Revit. The major barrier slowing down BIM implementation are the product suppliers and that includes, not only but to a great degree, Autodesk.

However i do agree plans, sections and elevations are dead, or nearly. Why draw in 3d only to convert back to 2d, it is just crazy and unrealistic to think we will continue doing this just because we have for so many years.

16 Dec 10 at 2:30 PM • Ian

Plans Elevations and sections are only as dead as our need for them may be. Where they are effective (comparativly) in transmitting the message necessary for successful project life cycle they will be used. The medium may be seen as the message to some but the agile choose the medium to suit the message.

Autodesk did have a toe in the door quite early but there were already many users expecting much more than that early vision and many would still not see that expectation as fullfilled. Lead foot doesn’t sound too far off the mark when it has taken a quarter century to get AutoCAD to partially implement the potential that the model has for containing, manipulating and broadcasting project lifecycle information.

Claiming ownership of the term is one thing but that is not ownership of the vision or the expectation.

AutoDesk’s biggest step involved buying Revit but it has taken a terribly long time for AutoCAD or Desktop to communicate adequately with that. They still do not.

In the mid-nineties AutoDesk joined others in IFC collaboration. IFC provides a way of usefully exchanging data between different elements in the related industries and between users of different tools. Absolutely essential to communication of information. Yet, a decade and a half on, Revit’s is a reasonable partial implementation, AutoCAD’s less so and Phil Bernstein is hung up on file format secrecy. This is a primary concern for AutoCAD; an over-self-protective attitude; a lump of lead.

File formats are minor players in the flow of information, the medium of exchange and the capacity to work them with are the users crown jewels.

Users are either not sufficiently demanding or are not of key interest to the developers.

When Phil Bernstein says that now, ‘nobody gives a shit’ who owns the data he means AutoCAD and perhaps other vendors do not give a shit. It does not improve the vendor’s results. The creator of the data DOES give a shit. We also see ownership of the model and the information it contains as a mechanism for effective coordination and dispersal of design intent and integrity. That ownership will remain vital to architect’s interests and we need to expect vendors to respond to that requirement.

I do agree, however that embracing more risk can actually reduce project risk. This is not an argument against ownership of the model but for it. Neither is it counter-intuitive. It has been the way of the successful masterbuilder and architect since Adam adjusted his fig leaf.

Finally, perhaps a jingoistic comment. Over a couple decades of particpation in architectural and CAD forii I have no reason to think the ‘if it ain’t broke don’t fix it’ dictum is more likely to rule individuals here than in America where it originated.

Phil Bernstein’s ideal that the uptake of technologies be accompanied by simultaneous deep investigation of the structural challenges to opportunity is one I strongly share and if one industry is proven to operate on the basis of that ideal better than another it is applaudable. The local industry does need to change circumstance to better value by more efficient technique (and so do many others around the world.)

17 Dec 10 at 5:41 AM • John Hainsworth

He’s spot on with his opinion sugessting we are “… trying to implement these technologies without having the broader discussion about what the process implications are”

These new tools are amazing, how we all change to apply them to deliver is what needs our attention.

His Archicad vs Revit point is akin to Rotriing vs Steadler. It doesn’t matter in the context of his argument – ignore it here please

22 Dec 10 at 3:54 PM • Anonymous

I think is quite interesting and completely agree that the days of elevations and plans are over, but just recently had an experience with this generation which Phil Bernstein describes, and while working entirely in 3d simulation and so forth when they were asked to provide plans and sections, you couldn’t fit a bed inside a bedroom and you couldn’t stand in the 1.50 spaces that they designed. So just basic intelligence will still be necessary.


Your email address will not be published.