The problem with using AEC software for 30 years is it’s long enough to see the same things over and over again too many times.

Everything said today in 2024, about ‘BIM’ was already said, verbatim, in 2004.

By 2004 there’d been enough time for leading A, E, and AE firms (architecture, engineering, and the combo) to develop sustained experience with BIM and its affiliated work process, predating the term ‘BIM’ already for 10 years for some. ArchiCAD was introduced already in 1987 and others around the same time or a bit later in the early 90s.

By 2004 there was enough experience for opinion formation. Many of those were complaints. Same as today.

So, yeah, 20 years ago we heard already all the same complaints we hear today. Like:

the software isn’t delivering the kind of outcomes that were promised

And back then, just as now, the answers were given, by software vendors, from ‘thought leaders’, from evangelists and enthusiasts and believers in change and so on. From conferences and papers, from white papers, from advocacy groups, from standards organizations. From many, in other words.

And the answers are the same now as they were then and often in the same words, word for word, the same 20 years ago as today:

The outcomes are not what was promised, but they would be if…

if teams established the right process. BIM is a process not a tool.

if users were given the right training.

if projects standardized on one software application.

if people would focus on ‘the I in BIM’. It’s the data, dummy!

if models conformed with data standardization schemas.

if the AEC industry weren’t the — multiple companies brought together on a project each looking out for its own interests — that it is.

if we could change contract law and even the premise of contract law — to something we can’t really define or put our finger on, but that’d be better for BIM process.

if not for the general condition, of the human psyche. We should seek to alter that to something better suited to BIM.

if not for slow-to-retire tech laggards who should make way faster for people with greater enthusiasm for the future and the software products defining it (the future).

There are more. I’ve forgotten some and left some out. I left out ‘blockchain’. For awhile we were told that blockchain will solve everything, by making transactions transparent.

But there were successful counter movements too. Generative design, for example, otherwise known as ‘computational design’, was already by 2004 well developed in some circles already using generative software for modeling. It didn’t solve everything though (of course). But it wasn’t supposed to. Many skilled people doing generative design work back then after awhile aimed their generative modeling at the problem of moving it as directly as possible to fabrication, to g-code. I keep respect for those who pioneered that software and persisted in its use and expansion.

But we should recognize the difference. Generative/computational design started from a tiny handful of just a few people focused on a clear target. They had a clear vision of where they were going and how to get there. And they did that. I think I’ve never heard complaints from any of those people. They saw what was missing, decided what to do, and did it, and followed through. No time for complaining. Just doing. Well, I’m oversimplifying. The complaining came before I met any of them or read what they wrote.

They complained probably that the system they wanted didn’t exist.

Then they built it.

I don’t maintain the same enthusiasm for BIM applications. I mean, I still like them. Sort of. Long ago I was obsessed with them. For years now I’ve been more interested in their limitations. And specifically, one fundamental limitation of AEC modeling software that I propose to solve.

I know what’s wrong with this software.

I mean, I know what’s missing that should not be missing. The fact that it is missing is self-defeating and counterproductive maximally. And I have a clear idea, of what to develop to solve that. What to add, to put back what’s missing.

What I don’t know is how to create a software company to do it. That’s (currently) beyond my skill set. If you have suggestions or you want to help build it with me, contact me on linkedin.

Here I’m gonna point again to my development proposal. First some background.

Long ago. I was involved in projects at Arch firms large and small. 20 years ago we had multi-discipline (A, S, MEP), multi-office, multi-firm digital design and construction modeling, with model sharing and coordination, automated drawing generation and re-generation, visualization apps, avatars in VR, some AR labs investigations, and generative/computational design flowing into some of those model pipelines. We had training. We had funding. We had management backing. We had standardization. And so on.

Here are some images of some of the projects I participated in back then:

Projects

Some of those are from a large firm where I was employed, and the small buildings are from a small firm later. They’re just mixed together on that roll as a sample.

In more detail, here’s a small project from 2007:

Model Automation and Model Quality on a Graph

Scroll down a bit on that page to see a video of the project model including A, S, M and P disciplines. Below the model are the model-automated drawings from A (architecture). Click on the caption link to view each drawing sheet full resolution. I’d done work with the same tools to the same degree of completion on several projects at another firm, in Louisville Kentucky already in 1999, now a quarter century ago.

In the video you see automated drawing-model fusion that was developed later by the software dev project I proposed and then led at Bentley Systems that automated the display of all the construction drawings in-situ within the model. More on that at the page here:

Earlier Innovations (2012)

I retrofitted the old project model from 2007 with the drawing-model fusion software upgrade in 2010 (while we were developing the softrware) and recorded the screen grab below (from MicroStation with the Arch add-on today called Open Buildings Designer that originated in the 90s as a Brics product called TriForma). The fusion found implementation in products from 9 different software companies since we did it first in 2012. But not in the dominant one.

To wrap up the background, look at this post from Simon Dilhas yesterday: https://www.linkedin.com/posts/simon-dilhas_can-we-blame-software-companies-for-wanting-activity-7167434812754980864-AHA_ Thanks for tagging me in the post, Simon. I tag back today. Read the comments under that linkedin post. Who can say the comments paint a great picture? It’s no surprise. Dissatisfaction with BIM software is well known.

Typical comments are about a general sloppiness of digital modeling in the industry and the decline in the quality of construction drawings. Complaints like that are often heard both from designers who’ve been around long enough to see the decline and from construction firms who are the sad recipients of degraded design deliverables,

On the comment thread others agreed:

1. there is a decline in quality of output from the design side
2. the decline has been seen over the last 20 years, and
3. it corresponds with the uptake of digital modeling tools over the same time period (I’m stopping short for now saying ‘is caused by’)

See this, from the Construction Brothers Podcast.

I’m not out on a limb saying something isn’t right. We’re stuck in a miasma. The 20 year old slogans explaining it away aren’t getting the job done anymore. They’re not believable. We’re sinking in a fetid swamp. So we look for a way out. How can we climb out of this, clean ourselves off, and walk with our feet dry one step ahead of the next going somewhere we want to go?

There’s something more tangible than the human psyche, or the nature of business itself, or the presence or lack of enthusiasm, training, or establishment among users of appropriate work process structuring very complex project teams and data flows… I mean, those are necessary and important but even when present and optimized something fundamental is still AWOL, its absence unauthorized so to speak. ‘86’d.’

I mean, absence of what’s missing is authorized, but it shouldn’t be. Here’s why:

Think for a second about the creation and use of highly elaborated models (mental and digital) of complex projects like buildings (small and large, and very large), bridges, power plants, and so on.

Models of these, mental and digital, don’t come into existence fully formed in an instant. Their development extends over long durations of time, months, years even.

Here’s a simplified encapsulation of some of the important factors in the development of complex mental models and their digital analog (as it were):

We all might reflect on the function of technical drawings, a function inseparable from models (mental and digital). The function is multifold:

1. Technical drawing is an expression of the act of looking somewhere specific within a model. The act of visual close study (VCS) and its articulation.  

2. There at that location, we evaluate, is everything that SHOULD be shown HERE actually shown here? Is anything that matters HERE missing?

3. Finally at some point after the long work of model development and review, someone with authority to do so, AFFIRMS the status of the questions in (2).

4. Along the way, an INTERPLAY is engaged between these many expressions of visual close study (VCS), which articulate the act of narrowed visual attentive focus, and the wider expansive environment of the whole of the project model. This interplay is a back and forth continuous dynamic. There is good argument that it -is- the basic observable dynamic of thought itself, that it IS a machine of thinking, an engine of thought. 

The idea that one side of the interplay can either be discarded, or stuck in a non-evolving centuries-old form of expression and externalized from the digital model, is simply self-defeating and counterproductive, Maximally. 

I draw your attention to item 4 in particular and emphasize it. For 30+ years now of marketing and development of digital modeling software in the AEC market, VCS has been left largely undeveloped.

This undermines the modeling endeavor itself. The engine of thought itself is left underpowered. 

This is a very significant omission from software capability, missing for more than 30 years now.

And it really is obvious:

If a company invents the automobile, unhitches the horse, but fails to design, manufacture, and install a motor… and blames others, that’s well and truly analogous to what AEC software corporations have given us.

Now come on. Think about this.

About both model quality, and drawing quality (and again, see this, from the Construction Brothers Podcast) — how many of the worst complaints would be well addressed and improved if software development put some attention again finally on VCS development, as a lens for looking at and evaluating model quality, and as a full half of the 2 cycle machine of thinking that IS the continuous perceptual loop of:

model << >> VCS INTERPLAY

At the site below is my proposal for open source development of VCS equipment for implementation in all AEC modeling (or model-handling) applications whose developers want their users to have it:

The Form of Engagement

Visual Engagement with Modeled Worlds

There is also a blog with my commentary https://tangerinefocus.com/blog-tangerine/

Anyone reading the proposal who wants to develop it, and wants my help working on it, please contact me on LinkedIn: https://www.linkedin.com/in/robsnyder3333/

TGN is for visual close study of digital models. It’s equipment for LOOKING, just as technical drawing, in VCS’s centuries-old conventional form, is for close study of mental models.

https://tangerinefocus.com

But what about AI?

Yeah, well, VCS matters as much as ever if not more, in models made with generative AI.

I foresee AI utility in three situations in which (VCS) visual close study equipment, in digital models of all kinds in AEC, plays a part in AI-assisted modeling and drawing work process:

1. The use of AI for both automatic placement of VCS (TGN) rigs at useful places throughout a digital model, and for automatic tuning of TGN settings within each of those rigs according to the logic of where they’re placed in the models.
   
2. ‘DeepQA’ question and answer against very complex physical asset (infrastructure) digital datasets may possibly be enhanced, giving better question and answer results, i.e., more productive, more useful, conversations with AI about difficult-to-answer hard questions, about very complex systems and huge volumes of data — because VCS (TGN) rigs for human interaction with, and visual close study of, those datasets over the long duration of design, construction, and operation of infrastructure assets MAY reveal data correlations that otherwise would be much harder (or impossible) to make, and, a richer field of correlation is a more fertile field for data mining and ‘AI’
   
3. Because generative AI generates digital models of AEC facilities IN AN ITERATIVE PROCESS, where a model is generated and then re-generated, again and again as the text prompt is tuned over and over again, VCS (TGN) rigs placed in these models (automatically as in item 1, or manually) for human visual close study of the model…, these VCS rigs are an optimal device within which to develop human-in-the-loop guidance mechanisms that allow the user to say, ‘OK, hold -these- parts of the generative model in -this- position/proportion/orientation/etc. and let the rest of the model re-iterate freely, unconstrained.