"CAD Worship"

[bnemec]

Anyone read the articles by Robert Green in Cadalyst? I find that many time's he hits it right on. I don't read all of them but I get some passed along. I think this two part article is worth linking to, I read them several years ago but stumbled across them again this morning, still very relevant:

https://www.cadalyst.com/management/avo ... sign-44187

edit: sorry, I had linked to the wrong part 2
https://www.cadalyst.com/management/avo ... rt-2-44455

[Tom G]

Good articles, thank you.
In other words, "Don't believe the hype."

[zxys001]

Anyone read the articles by Robert Green in Cadalyst? I find that many time's he hits it right on. I don't read all of them but I get some passed along. I think this two part article is worth linking to, I read them several years ago but stumbled across them again this morning, still very relevant:

https://www.cadalyst.com/management/avo ... sign-44187

edit: sorry, I had linked to the wrong part 2
https://www.cadalyst.com/management/avo ... rt-2-44455

Thanks.. not bad... he hit on a lot of good points.
Maybe at the end... "What cult are you allowing yourself to follow?"

[matt]

His main idea is good, but I think in his examples, he shows his own biases. Great title, but he should have left it at that. All the words he added detracted from the title.

[bnemec]

In other words, "Don't believe the hype."

Yeah, that's some of it. It's nothing personal against sales/marketing people, I just don't blindly trust someone that is trying to sell me something.

I think his point is that just using the flashiest tools doesn't make one a good designer. No different then if I were to buy the most hyped up fly rod it wouldn't change the fact that I cannot float a line in the air. Actually, I don't need the metaphor; having access to full Ansys, Catia and NX wouldn't make me a much better designer or engineer.

[bnemec]

His main idea is good, but I think in his examples, he shows his own biases. Great title, but he should have left it at that. All the words he added detracted from the title.

By "all the words he added" do you mean the article in it's entirety?

[matt]

By "all the words he added" do you mean the article in it's entirety?

His main premise is good, but his examples are one sided, and he seems to have a hard time seeing other points of view. The article seems to say "all those 3D salesmen are lying to you", when I'd rather he say something like "applying CAD tools to design work requires imagination, you have to figure out how to make it go where you want it to go".

Did you read his CAD Administration book? You just get the feeling that moving forward is hard and unnecessary work to him. 3D CAD is versatile. I've made a career using mid-range cad for stuff you're supposed to use high end cad for. People think Solid Edge is just about machine design, but you can do shapes and surfacing and mold work, and plastics, and a lot of stuff. There's not a single project I've done where I wished I was working in 2D to convey product information.

[bnemec]

His main premise is good, but his examples are one sided, and he seems to have a hard time seeing other points of view. The article seems to say "all those 3D salesmen are lying to you", when I'd rather he say something like "applying CAD tools to design work requires imagination, you have to figure out how to make it go where you want it to go".

Did you read his CAD Administration book? You just get the feeling that moving forward is hard and unnecessary work to him. 3D CAD is versatile. I've made a career using mid-range cad for stuff you're supposed to use high end cad for. People think Solid Edge is just about machine design, but you can do shapes and surfacing and mold work, and plastics, and a lot of stuff. There's not a single project I've done where I wished I was working in 2D to convey product information.

I agree with your points, even agree with he might have a hard time seeing other points of view.

I have not read his CAD Admin book.

I get the feeling that he's in the similar use case as the company I am at. Where moving forward IS very hard and is often fruitless in the end or a back slide of problems. Going back over the years the problems cause by features that were not 100% tested out before guys started using them are mind boggling and we were still dealing with them. Remember, there are a couple CAD users out there that maintain and copy CAD files for decades. I get the feeling that people that can run care free with all the newest wiz-bangs have a hard time seeing how much it costs when the new features fail down stream and cost hundreds or thousands fold of any potential increase in performance.

[matt]

I get the feeling that people that can run care free with all the newest wiz-bangs have a hard time seeing how much it costs when the new features fail down stream and cost hundreds or thousands fold of any potential increase in performance.

The thing I wonder about sometimes is why people who do machine design type work keep defending history-based design. You don't need it. History is just wacky overkill for machine design. Parts only have a few features each, and don't benefit from ordered features very often. Trying to keep data like that live for a decade or two is insane. It's probably crazy to keep it live even within a single release. Direct edit tools are what you need for machine design. They can deal with import files much better, and the intelligence is more in the software than in the data.

[bnemec]

The thing I wonder about sometimes is why people who do machine design type work keep defending history-based design. You don't need it. History is just wacky overkill for machine design. Parts only have a few features each, and don't benefit from ordered features very often. Trying to keep data like that live for a decade or two is insane. It's probably crazy to keep it live even within a single release. Direct edit tools are what you need for machine design. They can deal with import files much better, and the intelligence is more in the software than in the data.

I don't really know much about machine design. I'm assuming it's all one-offs? Design, build, forget (except for service parts maybe?) I assume they actually have a "standard parts library"

[zxys001]

The thing I wonder about sometimes is why people who do machine design type work keep defending history-based design. You don't need it. History is just wacky overkill for machine design. Parts only have a few features each, and don't benefit from ordered features very often. Trying to keep data like that live for a decade or two is insane. It's probably crazy to keep it live even within a single release. Direct edit tools are what you need for machine design. They can deal with import files much better, and the intelligence is more in the software than in the data.

yep, even though we've talked about this over/over/over many years... anyhow, you've nailed it.

[matt]

I don't really know much about machine design. I'm assuming it's all one-offs? Design, build, forget (except for service parts maybe?) I assume they actually have a "standard parts library"

Anything that's fairly simple shapes - prismatic shapes of stock purchased materials, and a lot of library parts.

[jcapriotti]

The thing I wonder about sometimes is why people who do machine design type work keep defending history-based design. You don't need it. History is just wacky overkill for machine design. Parts only have a few features each, and don't benefit from ordered features very often. Trying to keep data like that live for a decade or two is insane. It's probably crazy to keep it live even within a single release. Direct edit tools are what you need for machine design. They can deal with import files much better, and the intelligence is more in the software than in the data.

Not to defend it exactly, but we do machine and sheet metal designs and most parts are simple. Lots of brackets for sure. They stay live for decades. History based has never been an a big issue because they are simple. I open parts that are 20+ years old and there are no issues for the most part, again because they are simple. Some of the more complex castings or few surface models we have might have errors over time.

Now the assemblies are where things get the most wacky, I believe because they are big, 1000s of parts. Old assemblies might have mate errors because of the changes to mates over the years but most problems are due to the many user edits that occurred to the bottom level parts where the assemblies weren't checked and corrected. Face IDs lost because users couldn't figure out how to fix an error and deleted/recreated sketch geometry thus breaking a long chain of references. I know ST wouldn't have this issue......or would it?

[jcapriotti]

His main idea is good, but I think in his examples, he shows his own biases. Great title, but he should have left it at that. All the words he added detracted from the title.

This quote shows his bias.

They saw no efficiency gain. It turns out they were so efficient with their well-customized AutoCAD environment that even after months of effort, they were less efficient than when they started.

So it turns out that the CAD manager and power users who had built the customized AutoCAD environment over the years had done far more to support great design than those who decided to simply switch to a new tool. The right questions had not been asked, so the wrong path was chosen.

There is an element of truth to this and mirrors our environment. The problem is, it's not apples to apples and an unfair comparison. They are comparing a highly customized AutoCAD solution with out of the box SolidWorks. We had the same, and needed to customize SolidWorks as well. After that, we had the added benefit of SolidWorks to check for interferences and see problems that were still missed with the old "AutoCAD" process.

[bnemec]

This quote shows his bias.


There is an element of truth to this and mirrors our environment. The problem is, it's not apples to apples and an unfair comparison. They are comparing a highly customized AutoCAD solution with out of the box SolidWorks. We had the same, and needed to customize SolidWorks as well. After that, we had the added benefit of SolidWorks to check for interferences and see problems that were still missed with the old "AutoCAD" process.

Yeah. That example, although probably exists somewhere, is rare. Perhaps other examples would have been able to land the concept with a larger audience.

Thing is, I share some of the "bias" as it's being called that is exemplified here, and it's from not one, but multiple experiences. What is Bias called when it's substantiated with experiences?

[Tom G]

What is Bias called when it's substantiated with experiences?

Judgement

[jcapriotti]

Judgement

I'll see your "judgement" and raise you a "revelation",

[jcapriotti]

Yeah. That example, although probably exists somewhere, is rare. Perhaps other examples would have been able to land the concept with a larger audience.

Thing is, I share some of the "bias" as it's being called that is exemplified here, and it's from not one, but multiple experiences. What is Bias called when it's substantiated with experiences?

It is fair to say that 3d is not always the answer, which is what he was going for. AutoCad is a better tool for some types of work that require no 3d in the design. Wiring diagrams for example. Some mechanical design like a panel layout maybe, although we do those in 3d as well since we need a 3d model up higher.

[matt]

Judgement

To me, there's bias and then there's bias. Bias on it's own is not the worst thing. Everybody has some bias in some direction, so bias isn't the crime. The crime is not being able to see past your bias, or that you can't see anything but your bias. The world is a complex place. If you can only see it from one point of view, you miss a lot of things.

Green entitles his article "CAD Worship", probably meaning to take a rip at people who worship SolidWorks, but winds up being unintentionally ironic, and worshiping Autocad.

If you make gaskets, use 2D. Most of the rest of the world is 3D. I stopped having this argument about 20 years ago, that article is only 3 years old.

Now a lot of people could extend that argument to 3DX. The problem is that I'm not sure they have demonstrated a lot of benefits for 3DX yet. Simply being on the cloud in itself is not a benefit. You recognize the irony in a group that has broken away from the "progress" group complaining about an old school feller who won't progress.

[bnemec]

To me, there's bias and then there's bias. Bias on it's own is not the worst thing. Everybody has some bias in some direction, so bias isn't the crime. The crime is not being able to see past your bias, or that you can't see anything but your bias. The world is a complex place. If you can only see it from one point of view, you miss a lot of things.

Green entitles his article "CAD Worship", probably meaning to take a rip at people who worship SolidWorks, but winds up being unintentionally ironic, and worshiping Autocad.

If you make gaskets, use 2D. Most of the rest of the world is 3D. I stopped having this argument about 20 years ago, that article is only 3 years old.

Now a lot of people could extend that argument to 3DX. The problem is that I'm not sure they have demonstrated a lot of benefits for 3DX yet. Simply being on the cloud in itself is not a benefit. You recognize the irony in a group that has broken away from the "progress" group complaining about an old school feller who won't progress.

I see what you mean with bias vs Bias as in tendency vs prejudice. I'm guilty of the "cannot see past it" sometimes (hopefully not always) and try to look for truth. Learning, slowly, that my hang-up is usually a bad assumption on my part that others are not plagued with. Then someone comes along and recognizes it and is kind enough to point it out and the lights come on.

I might be misunderstanding his point. I was taking it as the tool, in and of itself, doesn't make people better or more proficient. It certainly can and usually does but it requires a lot more than just the purchase and use of the tool; training on the new tool and new processes around the new tool and training on the new processes... all play just as big of a role.

[HerrTick]

probably meaning to take a rip at people who worship SolidWorks

SW has (had?) a very loyal and dedicated community. But we collectively were hardly devout.

Old school PTC users, on the other hand...

[zwei]

In reality, here’s what happened when they switched over:

They already had utilities that created BOMs and assembly drawings in AutoCAD, and they now had to start over learning how those processes worked in SolidWorks.
Shop floor tools such as viewing utilities, NC software, drill machine interfaces, etc. didn’t work as expected anymore.
Users required extensive training on SolidWorks, and they were working more hours to do the same tasks in new ways.

Personally i feel this example is really ironic...
If you are switching a system entirely, of course you will need to start over learning, same goes with your other software that work with your CAD system...
Expecting users to know how to do everything the moment you switch without any training is just... wrong...

It is just a matter of whether the time for training and re-setting up the system justify the benefit of switching.

Take the SWYMP as example, lots of us is here because we don't see the benefit of training ourselves to be a SWYMP master as there is literally no benefit there.

[SamSpade]

There are some truths to the articles, and thanks again for posting them. I understand the writer's bias, but I don't necessarily agree with all of his analysis. The analogy that you don’t need to kill a fly with a sledgehammer when a fly swatter will do, does hold true to some extent. On the other hand, the argument of why fix something if it isn’t broken, is simply wrong. I never understood this reasoning. Yes, the status quo maybe good, and may in effect yet be the best present-day solution out there, but shouldn’t we continually strive to see if there is a better way to do things. Scientific breakthroughs and achievements happen because someone decides that we can do and should do better. Evolution and progress are inevitable, as are the growing pains that come along with them.

Last year I was tasked to prepare a report on how to best use the two CAD programs in house (AutoCAD & SWX). Basically, to determine if completely switching over to SWX for all manufacturing purposes was the way to go (which was my initial bias leaning – more continuity in manufacturing). Both CAD programs have been in use here for the last 20 years, prior to that it was solely 2D. After speaking with colleagues and considering their proficiency in both CAD systems, I came to the conclusion that both were still necessary. It basically came down to aptitude, or the lack there of, and the efficacy of certain AutoCAD elements. (Due to the nature of our industry, architectural, which is still heavily AutoCAD based, for the moment, 2D software is still necessary for manipulating shop and architectural drawings.)

Over the years our department has developed many 2D AutoCAD templates, that contain complex dynamic blocks, for standard product types where the number of variables is known. For example, a simple curved metal ‘C’-shaped wall molding. The unknown variables are; direction of curve (in or out), material type, finish type, thickness, width, height, length, and radius of molding. We have an AutoCAD template that allows the end user to input these variables (attributes) and the template in turn generates a fully dimensioned drawing, along with the corresponding DXF flat development for exporting. There is no drawing layout manipulation required as the curved view of the molding is intentionally drawn not to scale, and there is very limited title block requirements because most fields are automatically populated. It is simple, accurate and quick. Can this this template be reproduced in SWX? Yes, definitely, with two file types to maintain (sldprt & slddrw) and most likely with drawing view manipulation being necessary. In this case the 2D approach is still preferred.

I guess it comes down to the type of work being done, and knowing which tool is the ideal tool for which job.

[jcapriotti]

@SamSpade Our industry is much the same. Simple, repetitive, minor tweaks to a known design within certain ranges just don't require any 3d. In some cases they have gone even simpler than AutoCAD and have the drawings in a PDF package with some calculations and macros to update quantities. One PDF file containing several drawings and BOMs. Speed is super important and as long as they are within the ranges and limits of the design, it works fine.