Big data and machine learning are disrupting industries left and right, making them exponentially more efficient and effective. However, when it comes to AEC specifically architecture, it seems like its five years behind. Part of me believes it has to do with a lot with liability, legacy of old paradigmatic thinking and fear. Watching lectures from Koolhaas and other major figures there is a constant skepticism and downplay of the ability of the computer programmer to affect change on the urban scale. Rather than learn from other industries/startups (for example startups using machine learning to create real time cartography) and leap frog this technology into our own workflows, it seems the older generation is largely in denial.
With that said, there are without a doubt firms such as NBBJ, SOM, Arup, etc pushing quickly in the oppposite direction but it seems technological adoption is slow. I personally find that firms that consist of only architects fail to adapt quickly. It is the programmer or data scientist that can make large datasets actionable for the architect. The BIM manager/computational designer who can streamline tools and how we execute tasks. We are definitely seeing a shift, but it seems like only large firms with large human resources are the only ones with enough foresight to affect change. This is changing as students today have this fundamental understanding. To me it seems necessary that firms (and more so individual architects) to be versed in scripting and to understand how to manage data sets to move from intuition based guessing to data driven designers. Thoughts or observations?
Non Sequitur
Feb 27, 17 4:28 pm
Because at the end of the day, you still need to give clear instructions to the guy swinging the hammer.
Also, buildings take several years to get out of the design phase and into construction hence the 5y delay.
chigurh
Feb 27, 17 5:46 pm
data is boring, poetics is not.
Scripting is useful if you need to automate a repetitive task, but there is a huge disconnect between what you are calling managing data sets and and artful building. Where does this data com from, how does it yield meaningful form, how many architects would even know how to analyze the data set or have the skill set to even do so. Computers are good at optimizing. Do you think that a building like Ronchamp would ever be the result of a computer program analyzing data? I don't.
Dangermouse
Feb 27, 17 8:10 pm
tech adoption is low in AEC because margins are low, projects are long, and there are no financial payoffs for "actionable big data". not for any of the reasons you listed. rem is skeptical about big data because he is correct, it is largely useless outside of a few niche applications/projects smaller than a certain scale.
shellarchitect
Feb 27, 17 8:28 pm
Which industries are becoming exponentially more efficient? Last I checked productivity had been going down for a while now
archi_dude
Feb 27, 17 9:32 pm
The big ideas, adoption of technology and streamlining of deliverables is coming from the construction firms. Architecture firm's services / workflow hasn't adapted in geez...forever? I'd say it's a combo of the architecture culture that views the words "efficiency" "business" and "profits" as evil attempting to take root in a pure profession based in the arts and also the fact that we can't hide behind a LLC or corporation for our mistakes like contractors and developers can.
bikebicycle
Feb 27, 17 10:28 pm
Big data and machine learning are disrupting industries left and right, making them exponentially more efficient and effective.
Your first statement seems rather specious. Do you have examples of how machine learning and big data are "disrupting" whole industries?
I'd really like to understand how what you are talking about is anything more than the fact that the software is better and computers are more powerful and can hold more data than they could 20+ years ago when they first "disrupted" this industry.
dominiond
Feb 27, 17 10:57 pm
Agree with OP. The AEC industry works in the Dark Ages. We should look to the automotive industry as an exemplar of more nimble design technologies and production.
I've worked on $100 million buildings that are outfitted with technology or electrical systems that are about to outmoded or at capacity by their opening. It's like when you go to a hotel and they have IPhone 3 chargers at the bedside table but at whole building scale. The entire construction process is generally too cumbersome to do a pivot and adapt to new and better technology.
Most architects (practicing and academic) spend too much time talking to themselves (like this entire forum) while thinking their ideas are the best- unfortunately, their training makes them that way. The really interesting design work is being done by software engineers, civil engineers, scientists and artists.
As far as coding, I'm teaching myself Python for fun and I encourage my studio to do so as well. I know that in 3 years, robots are going to be doing the majority of coding, but I figure I should arm myself with some rudimentary tools so I'm at least considered for when the singularity happens.
DTL DWG
Feb 27, 17 11:01 pm
generally speaking a hand sketch will take about 3 minutes and explain about 10GB worth of data and can be sent via email....a 2D CAD drawing could take 15 minutes to an hour and explain about 20 GB of data....so there you go - representation!
in my daily routine I have hardly any use for the following - parametricism, 3d prototyping, very fancy theories on engineering...like 99.9% of the time, absolutely no use and would only loose money worrying about them, I like making money and being an architect, my bad....
your conscience
Feb 27, 17 11:05 pm
The industry cannot be any further ahead than it is, at least in this timeline. The problems holding back the industry have nothing to do with technology but with trust, job security, personal inferiority complexes and communication issues.
DTL DWG
Feb 27, 17 11:08 pm
an example of representation:
I expect the data on this by tomorrow morning. thanks.
think_again
Feb 27, 17 11:43 pm
Here's the full data:
JFIF
JFIF Version 1.01
Resolution 1 pixels/None
File — basic information derived from the file.
File Type JPEG
MIME Type image/jpeg
Comment CREATOR: gd-jpeg v1.0 (using IJG JPEG v62), quality = 90
Encoding Process Baseline DCT, Huffman coding
Bits Per Sample 8
Color Components 3
File Size 144 kB
Image Size 1,008 × 1,077
Y Cb Cr Sub Sampling YCbCr4:2:0 (2 2)
Jose J Pullutasig
Feb 27, 17 11:53 pm
We never moved beyond pen and paper. Parametric modeling was introduced in 1963. Autocad added the feature in 2009. It took 49 years for parametrics to clutter the cad menu.
Although C. Price, C. Alexander and N. Negroponte thrived in the field of cybernetics, robotics and artificial intelligence.
bikebicycle
Feb 28, 17 1:21 am
to be fair - the AEC industry is pretty far behind disruptive big data and machine learning tech in creating fake markets that subsist primarily on venture capital.
Volunteer
Feb 28, 17 8:04 am
Well, here is the Pons Fabricius bridge in Rome that was built in 62 BC and has been in constant use since. Not real sure what computers would have added in this case.
curtkram
Feb 28, 17 9:14 am
did you get to work on that project volunteer?
cipyboy
Feb 28, 17 9:29 am
"...Koolhaas and other major figures there is a constant skepticism and downplay of the ability of the computer programmer to affect change on the urban scale"
First, Rem didn't strike me as someone who does not advocate technology- I have watched his lectures, Eisenman maybe yes. Second, his thinking in architecture goes far beyond computer and technology. For every additional line or curve means added resources and skills- we do have a finite number of builders who can actually build decent buildings, let alone produce innovative buildings.
Volunteer
Feb 28, 17 10:56 am
Curtkram, Didn't quite make that one. Here is a pix showing the center arch coming into play at a flood stage of the Tiber River. Reduces stress on the bridge and allows more water downstream Simple, elegant.
techtonic
Feb 28, 17 12:56 pm
I understand there is still merit in the analogue modes of design. There are things algorithms won't get right. I think the debate of whether or not this technology is useful to architects reflects that much of this profession still sees the adoption of these tools as either pointless or even harmful to the profession. Not to be ageist, but perhaps I am speaking to a more established, knowledgeable generation of practitioners that have experienced the profession under different circumstances and have a difficult time seeing the profession become more automated.
I think dangermouse's point makes a lot of sense: that the AEC industry does not have large VC firms backing innovation that allows companies to be non-profitable for years before doing so. I wonder if there are examples of architecture firms being funded this way? The only example I can think of is NBBJ venture backing Visual Vocal. This paradigm does have its downfalls (major bubbles) but it does seem to be one that allows for more R&D and thus more innovation, something smaller firms often don't have a budget for.
To respond to bikebicycle, there are several industries leveraging the use of data aggregation to transform themselves from conventional industry to conventional industry + data analytics. An example is the agriculture industry. These tools are making industries more prescriptive in use and deployment of resources and I see how the architecture/construction industries share many of the timelines as agricultural/automotive production do but have not utilized abundant computational tech in streamlining its processes. This is also a good read.
Machine learning to create real time cartograhy is not jargon but a description of what is happening in the driverless vehicle/LIDAR space. Example. If cars (read moving spaces) can communicate with each other in real time to coordinate resources (occupied road) why can't we leverage mining the data buildings hold to better understand how we occupy and use our architecture, instead of relying on intuition or heuristics to fill the gaps?
dominiond, I am too learning python but must do so on my time. I wonder if you see merit in allowing employees to spend a few hours a week learning this? I understand its hard to see a clear ROI but I also think firms that aren't promoting active learning and adoption of new skills are shooting themselves in the foot in the long run. I'd like to know some of your thoughts on budgeting for continuing education.
As always, thanks for the thoughtful responses.
mightyaa
Feb 28, 17 3:17 pm
I’ll play. I need some examples of how you believe it would add value to project. It’s one thing to talk about cars communicating to each other and traffic control systems to achieve a single goal (traffic moving along smoothly). There are huge advantages for that; People are already using the roads and you can gather useful data. All you are really doing is optimizing what already is; that isn’t true change.
Data measures something. “If cars (read moving spaces) can communicate with each other in real time to coordinate resources (occupied road) why can't we leverage mining the data buildings hold to better understand how we occupy and use our architecture, instead of relying on intuition or heuristics to fill the gaps?” The physics of a moving vehicle has several variables that can be adjusted on the fly dynamically. Buildings are static. We do use these sorts of systems for the variable elements; sun control, HVAC are two good examples. Smaller ones tend to kick in with scale; Advanced elevator controls to reduce wait times and anticipate loads, vehicle parking systems, etc. But no matter what you do, a hallway width won’t change.
If you are talking a much broader holistic approach… Market assessment reports are pretty normal before committing multi-millions on a speculative development. Basically, no one builds on a whim. Many are also smart and know a really good architect can attract people through design. That’s where the other breaks apart; What we do through architecture is illicit an emotional response as much as satisfy a need. You can build the 200 living units the data suggests should be ‘here’, but your data can’t sort out the complex emotional response for why someone would want to… “How they feel about the idea of living in that particular space.”
To circle it around to your car example: You solved the wrong problem. That data won't help in deciding why they are in "this car" instead of "that car". The guy selling cars and spending billions wants you in his car; he doesn't give a rat's arse about anything that doesn't play into that.
x-jla
Feb 28, 17 3:41 pm
Ive always been boggled as to why architects rarely if ever design building products/systems. Thats where innovation has the most potential...in the parts and systems themselves...not the assemblage.
Peter Normand
Feb 28, 17 9:19 pm
Jla-X Lots of architects have a hand in building products. they may not get on the cover of the magazines but architects are often part of the team designing or refining building products. Furniture and lighting is also heavily influenced by design talent from our industry.
The notion that Architecture has a Luddite tendency as a profession is not new. I argue that given the financial and environmental cost of buildings it is important to mitigate risk. Most buildings are singular designed objects. There are few options for automation of design and construction because each building has it's own shape and unique site conditions. The advantages of automation occur where you are repeatedly doing the same task in exactly the same circumstances with the same result as the intended outcome. It is hard to look at architecture, even the cheap commodity like investment architecture, as having very many opportunities for the investment in automation to pay off without fundamentally cheapening what may believe is one of the most interesting and worth while aspects of architecture, uniqueness.
People place value on having some semblance of uniqueness and that becomes hard to do in an automated design or construction system.
The other reason the architecture and construction industry is not likely to benefit from much more data driven automation than we already have is the huge and constantly shifting number of variables. Agriculture has a lot of variables but they get to repeat the same experiment with each harvest, Architects if we do our jobs well don't get an annual do over and thus piles of data to mine that would be relevant to the next round. Automation has to contend with a lot of variables if it were to take on the design of a building, many of those variables are not easy if not impossible to measure as they are more emotional and subjective.
Many people, myself included, believe we can see and intuit the difference in something that is engineered and something that is designed, and that the designed object often is preferable to one that is engineered because design has value and design can not be easily reduced to a set of data variables. Emotion and art, even if it is an applied art like architecture, can not and should not be automated.
Big data and machine learning are disrupting industries left and right, making them exponentially more efficient and effective. However, when it comes to AEC specifically architecture, it seems like its five years behind. Part of me believes it has to do with a lot with liability, legacy of old paradigmatic thinking and fear. Watching lectures from Koolhaas and other major figures there is a constant skepticism and downplay of the ability of the computer programmer to affect change on the urban scale. Rather than learn from other industries/startups (for example startups using machine learning to create real time cartography) and leap frog this technology into our own workflows, it seems the older generation is largely in denial.
With that said, there are without a doubt firms such as NBBJ, SOM, Arup, etc pushing quickly in the oppposite direction but it seems technological adoption is slow. I personally find that firms that consist of only architects fail to adapt quickly. It is the programmer or data scientist that can make large datasets actionable for the architect. The BIM manager/computational designer who can streamline tools and how we execute tasks. We are definitely seeing a shift, but it seems like only large firms with large human resources are the only ones with enough foresight to affect change. This is changing as students today have this fundamental understanding. To me it seems necessary that firms (and more so individual architects) to be versed in scripting and to understand how to manage data sets to move from intuition based guessing to data driven designers. Thoughts or observations?
Because at the end of the day, you still need to give clear instructions to the guy swinging the hammer.
Also, buildings take several years to get out of the design phase and into construction hence the 5y delay.
data is boring, poetics is not.
Scripting is useful if you need to automate a repetitive task, but there is a huge disconnect between what you are calling managing data sets and and artful building. Where does this data com from, how does it yield meaningful form, how many architects would even know how to analyze the data set or have the skill set to even do so. Computers are good at optimizing. Do you think that a building like Ronchamp would ever be the result of a computer program analyzing data? I don't.
tech adoption is low in AEC because margins are low, projects are long, and there are no financial payoffs for "actionable big data". not for any of the reasons you listed. rem is skeptical about big data because he is correct, it is largely useless outside of a few niche applications/projects smaller than a certain scale.
Which industries are becoming exponentially more efficient? Last I checked productivity had been going down for a while now
The big ideas, adoption of technology and streamlining of deliverables is coming from the construction firms. Architecture firm's services / workflow hasn't adapted in geez...forever? I'd say it's a combo of the architecture culture that views the words "efficiency" "business" and "profits" as evil attempting to take root in a pure profession based in the arts and also the fact that we can't hide behind a LLC or corporation for our mistakes like contractors and developers can.
Big data and machine learning are disrupting industries left and right, making them exponentially more efficient and effective.
Your first statement seems rather specious. Do you have examples of how machine learning and big data are "disrupting" whole industries?
machine learning to create real time cartography
What does this even mean? Are you using jargon to refer to generating forms in 3d software, or are you talking about something more akin to using algorithms that take GIS and demographic data and blindly deny certain neighborhoods access to services.
I'd really like to understand how what you are talking about is anything more than the fact that the software is better and computers are more powerful and can hold more data than they could 20+ years ago when they first "disrupted" this industry.
Agree with OP. The AEC industry works in the Dark Ages. We should look to the automotive industry as an exemplar of more nimble design technologies and production.
I've worked on $100 million buildings that are outfitted with technology or electrical systems that are about to outmoded or at capacity by their opening. It's like when you go to a hotel and they have IPhone 3 chargers at the bedside table but at whole building scale. The entire construction process is generally too cumbersome to do a pivot and adapt to new and better technology.
Most architects (practicing and academic) spend too much time talking to themselves (like this entire forum) while thinking their ideas are the best- unfortunately, their training makes them that way. The really interesting design work is being done by software engineers, civil engineers, scientists and artists.
As far as coding, I'm teaching myself Python for fun and I encourage my studio to do so as well. I know that in 3 years, robots are going to be doing the majority of coding, but I figure I should arm myself with some rudimentary tools so I'm at least considered for when the singularity happens.
generally speaking a hand sketch will take about 3 minutes and explain about 10GB worth of data and can be sent via email....a 2D CAD drawing could take 15 minutes to an hour and explain about 20 GB of data....so there you go - representation!
try this article - A City Is Not a Computer
in my daily routine I have hardly any use for the following - parametricism, 3d prototyping, very fancy theories on engineering...like 99.9% of the time, absolutely no use and would only loose money worrying about them, I like making money and being an architect, my bad....
The industry cannot be any further ahead than it is, at least in this timeline. The problems holding back the industry have nothing to do with technology but with trust, job security, personal inferiority complexes and communication issues.
an example of representation:
I expect the data on this by tomorrow morning. thanks.
Here's the full data: JFIF JFIF Version 1.01 Resolution 1 pixels/None File — basic information derived from the file. File Type JPEG MIME Type image/jpeg Comment CREATOR: gd-jpeg v1.0 (using IJG JPEG v62), quality = 90 Encoding Process Baseline DCT, Huffman coding Bits Per Sample 8 Color Components 3 File Size 144 kB Image Size 1,008 × 1,077 Y Cb Cr Sub Sampling YCbCr4:2:0 (2 2)
We never moved beyond pen and paper. Parametric modeling was introduced in 1963. Autocad added the feature in 2009. It took 49 years for parametrics to clutter the cad menu.
Although C. Price, C. Alexander and N. Negroponte thrived in the field of cybernetics, robotics and artificial intelligence.
to be fair - the AEC industry is pretty far behind disruptive big data and machine learning tech in creating fake markets that subsist primarily on venture capital.
Well, here is the Pons Fabricius bridge in Rome that was built in 62 BC and has been in constant use since. Not real sure what computers would have added in this case.
did you get to work on that project volunteer?
"...Koolhaas and other major figures there is a constant skepticism and downplay of the ability of the computer programmer to affect change on the urban scale"
First, Rem didn't strike me as someone who does not advocate technology- I have watched his lectures, Eisenman maybe yes. Second, his thinking in architecture goes far beyond computer and technology. For every additional line or curve means added resources and skills- we do have a finite number of builders who can actually build decent buildings, let alone produce innovative buildings.
Curtkram, Didn't quite make that one. Here is a pix showing the center arch coming into play at a flood stage of the Tiber River. Reduces stress on the bridge and allows more water downstream Simple, elegant.
I understand there is still merit in the analogue modes of design. There are things algorithms won't get right. I think the debate of whether or not this technology is useful to architects reflects that much of this profession still sees the adoption of these tools as either pointless or even harmful to the profession. Not to be ageist, but perhaps I am speaking to a more established, knowledgeable generation of practitioners that have experienced the profession under different circumstances and have a difficult time seeing the profession become more automated.
I think dangermouse's point makes a lot of sense: that the AEC industry does not have large VC firms backing innovation that allows companies to be non-profitable for years before doing so. I wonder if there are examples of architecture firms being funded this way? The only example I can think of is NBBJ venture backing Visual Vocal. This paradigm does have its downfalls (major bubbles) but it does seem to be one that allows for more R&D and thus more innovation, something smaller firms often don't have a budget for.
To respond to bikebicycle, there are several industries leveraging the use of data aggregation to transform themselves from conventional industry to conventional industry + data analytics. An example is the agriculture industry. These tools are making industries more prescriptive in use and deployment of resources and I see how the architecture/construction industries share many of the timelines as agricultural/automotive production do but have not utilized abundant computational tech in streamlining its processes. This is also a good read.
Machine learning to create real time cartograhy is not jargon but a description of what is happening in the driverless vehicle/LIDAR space. Example. If cars (read moving spaces) can communicate with each other in real time to coordinate resources (occupied road) why can't we leverage mining the data buildings hold to better understand how we occupy and use our architecture, instead of relying on intuition or heuristics to fill the gaps?
dominiond, I am too learning python but must do so on my time. I wonder if you see merit in allowing employees to spend a few hours a week learning this? I understand its hard to see a clear ROI but I also think firms that aren't promoting active learning and adoption of new skills are shooting themselves in the foot in the long run. I'd like to know some of your thoughts on budgeting for continuing education.
As always, thanks for the thoughtful responses.
I’ll play. I need some examples of how you believe it would add value to project. It’s one thing to talk about cars communicating to each other and traffic control systems to achieve a single goal (traffic moving along smoothly). There are huge advantages for that; People are already using the roads and you can gather useful data. All you are really doing is optimizing what already is; that isn’t true change.
Data measures something. “If cars (read moving spaces) can communicate with each other in real time to coordinate resources (occupied road) why can't we leverage mining the data buildings hold to better understand how we occupy and use our architecture, instead of relying on intuition or heuristics to fill the gaps?” The physics of a moving vehicle has several variables that can be adjusted on the fly dynamically. Buildings are static. We do use these sorts of systems for the variable elements; sun control, HVAC are two good examples. Smaller ones tend to kick in with scale; Advanced elevator controls to reduce wait times and anticipate loads, vehicle parking systems, etc. But no matter what you do, a hallway width won’t change.
If you are talking a much broader holistic approach… Market assessment reports are pretty normal before committing multi-millions on a speculative development. Basically, no one builds on a whim. Many are also smart and know a really good architect can attract people through design. That’s where the other breaks apart; What we do through architecture is illicit an emotional response as much as satisfy a need. You can build the 200 living units the data suggests should be ‘here’, but your data can’t sort out the complex emotional response for why someone would want to… “How they feel about the idea of living in that particular space.”
To circle it around to your car example: You solved the wrong problem. That data won't help in deciding why they are in "this car" instead of "that car". The guy selling cars and spending billions wants you in his car; he doesn't give a rat's arse about anything that doesn't play into that.
Ive always been boggled as to why architects rarely if ever design building products/systems. Thats where innovation has the most potential...in the parts and systems themselves...not the assemblage.
Jla-X Lots of architects have a hand in building products. they may not get on the cover of the magazines but architects are often part of the team designing or refining building products. Furniture and lighting is also heavily influenced by design talent from our industry.
The notion that Architecture has a Luddite tendency as a profession is not new. I argue that given the financial and environmental cost of buildings it is important to mitigate risk. Most buildings are singular designed objects. There are few options for automation of design and construction because each building has it's own shape and unique site conditions. The advantages of automation occur where you are repeatedly doing the same task in exactly the same circumstances with the same result as the intended outcome. It is hard to look at architecture, even the cheap commodity like investment architecture, as having very many opportunities for the investment in automation to pay off without fundamentally cheapening what may believe is one of the most interesting and worth while aspects of architecture, uniqueness.
People place value on having some semblance of uniqueness and that becomes hard to do in an automated design or construction system.
The other reason the architecture and construction industry is not likely to benefit from much more data driven automation than we already have is the huge and constantly shifting number of variables. Agriculture has a lot of variables but they get to repeat the same experiment with each harvest, Architects if we do our jobs well don't get an annual do over and thus piles of data to mine that would be relevant to the next round. Automation has to contend with a lot of variables if it were to take on the design of a building, many of those variables are not easy if not impossible to measure as they are more emotional and subjective.
Many people, myself included, believe we can see and intuit the difference in something that is engineered and something that is designed, and that the designed object often is preferable to one that is engineered because design has value and design can not be easily reduced to a set of data variables. Emotion and art, even if it is an applied art like architecture, can not and should not be automated.
Over and OUT
Peter N