[onthecanposting]

Construction technology. Design software for horizontal construction is astonishingly bad. The PDF is still the "product" in am era where GPS and automated machine guidance are the norm. We're decades overdue on rethinking project delivery from first principles in the context of modern practices and The inefficiency costs billions. A single bill or FHWA policy memo on open data standards for digital project delivery will be the starting gun for an explosion of disruption in this market. No telling who will win, but it won't be Bentley and I doubt Autodesk can innovate fast enough.

[kyriakos]

I can't predict what the future will bring in this sector but I'm in the process of building a house and had to undergo through the architectural process and planning (by hiring professionals of course not going at it on my own). I am surprised of how advanced things are with building construction, eg with realtime rendering, unreal engine 5 integration into autodesk revit, actually making changes and being instantly visible with photorealism feels like science fiction. All quantities for materials are calculated automatically which makes cost estimations much more accurate and allows you to play with options and iterate designs in a way that wasn't possible before.

Of course construction is not only for buildings but for someone who had no experience in this sector I was caught off guard expecting to see old fashioned blue prints.

[teddyh]

All that could very well be true. The important question is: Is there anyone who is prepared to spend money to fix any of this?

Put another way: It doesn’t matter if you have a product which would help people and save them money, if people, for whatever reason, aren’t inclined to buy your product.

[onthecanposting]

Bentley has a market cap of around $15B. It's worth something to somebody. Their only serious market is DOT work and their position is attributable to administrative inertia and aggressive vendor lock-in.

The biggest problem isn't making better software, which already exists, but getting DOTs to stop making Bentley's proprietary file type a requirement for deliverables. The second problem is realizing that the CAD approach of imitating a paper process from last century isn't actually necessary.

[m0llusk]

Current owners, operators, and tenants. Buildings need to work for the occupants. Even in the best scenarios this takes ongoing upgrades and maintenance. In some cases buildings need to be significantly designed to provide for the needs of current tenants. This means that the whole construction sector needs the hardware equivalent of continuous integration based on the idea that construction is only the start of a process of keeping a building operational for its service lifetime.

[teddyh]

It’s the same classic problem when selling business products; the buyers are usually not the users, so no user-friendly features are created; only features which would look good in a demo to the buyers.

Therefore, no products which only contain features which help users (and save them money) can be a success, since the buyers won’t be impressed by it, and will not pay for it.

[aothms]

It's also quite fun! There's a lot of computational geometry involved and you have a chance to positively affect the built environment around you.

We're quite close now to delivering the final version of IFC4.3: the open standard for BIM (building information modelling), now including semantic modelling of long linear infra https://ifc43-docs.standards.buildingsmart.org/ Eagerly awaiting to replace your PDF.

[onthecanposting]

Since you are an insider, would you please shed some light on a personal question I've had for a while? Is the reason the IFC specification is so extremely complicated because they're trying to map nicely to an object model that's already in Autodesk products?

I read a proposal a while ago to simplify an open BIM spec down to meshes for solids and a generic dictionary for attaching property data. That struck me as very sane and very easy to implement. You could have a free product like Blender spit out BIM compliant deliverable at minimal cost. Apparently some high ups at Autodesk talked to the author and he came to understand the error of his ways.

[aothms]

This is a difficult question and the the history to all of this predates my involvement.

Indeed, a fair amount of people believe that the IFC is intentionally unnecessarily complex to limit interoperability and retain a monopoly.

Over time, I've come to believe hanlon's razor is also in play here and it's more of a poor understanding of use cases, academic ideas blowing up scope and inheriting obscure schema and serialization tech.

Meshes + metadata can facilitate most of the use cases in industry today, which is coordination, interference checks and visualization.

But at the same time, there's some pretty compelling use of the standard that requires slightly more semantic geometry descriptions:

- calculating geometric quantities according to local norms, which requires some additional geometric knowledge on things like openings and the axis of a wall for example

- The opening direction of doors is often good to know, but this is just a convention on the local transformation matrix, so could be just meshes.

- steel manufacturing can derive quite a bit of information from the IFC geometry, like parametric cross section profiles and where to drill for bolt holes and using which diameter

And then there is the ultimate end goal according to some to be able to exchange all parametric and constraint information from the native model. But this is still quite far out.

The challenge for future editions of the spec is to better align required complexity of certain use cases with a more modular spec so that you depend on a more appropriate amount of complexity.

Btw. since you mention Blender. The BlenderBIM addon https://blenderbim.org/ is actually one of the most avid users of complex and parametric constructs in IFC :)

[onthecanposting]

IFCOpenShell was news to me. Thanks for sharing. You're doing God's work.