My obsession for/against various CAD/PLM tools

Those who know me for a while know about my somewhat erratic career path - from hands-on mechanical design in Military/Defense, through a stint of few years in Software (CAD) product management, then back to hardcore mechanical engineering. One of the byproducts of this odd route is having been exposed to most (all?) popular CAD+PLM packages. And I have, uhm... strong but informed opinions.

From my early years as CAD user I was already pretty vocal about my grievances, and I had the opportunity to voice them unfiltered, directly to some of the people who were making these tools. I'm sure Eyal Chaki is still healing from the trauma first meeting me on "user feedback" sessions I attended many years ago :)

More recently, I've been having some "fiery but mostly peaceful" debates (verging trolling? sorry, sometimes I get carried away) with Oleg Shilovitsky (a PLM "thought leader" to an extent) about the nature, history, trends and prospects of various changes in the industry, where I often express my very rigid skepticism and negative stance against any new tools, in a market that is already flooded with productivity-killing swaths of software packages, that live in the periphery of CAD, ever-expanding, ever-adding more stuff for design engineers - like me - to waste our time "managing" data that is often a duplication of what we already do either directly in CAD, or in some Excel sheet that needs no "managing". I consider that as a service to the majority of design engineers that suffer silently day in and day out, interacting with bad PLMs with no choice. Y'all are welcome.

A side-effect of my history is that many in my direct and secondary network are in the CAD/PLM industry, which often leads to discussions and inquiries about my opinion on some tools, and sometimes solicitation of interest in new tools, assuming I'm a potential "lead" for sales (I'm not). But sometimes I respond and offer my honest, and usually harsh take: Nobody needs any more PLM tools like nobody needs a single more government office created.

So I figured, maybe worth putting this write-up out there, hoping it will better clarify my perspective for past and future debates. Also - it's not a debate. I'm 100% right. Most of the time... ;)

(Warning: some political views might be popping up here and there)

Bottom Line Up Front: All PLM tools are productivity killers.

It's 2024. Here's one real (redacted) screenshot that tells you all you need to know about how bad PLM tools are:

On the left - you see a very simple CAD assembly structure of two plain ol' parts. On the right - you see the exact same assembly of the same two parts, the way PLM shows them. WHY?!

Why does a design engineer HAVE to see the entire collection of any and all auxiliary files, viewables, meta-data, linked data, layers upon layers of stuff that NOBODY CARES ABOUT. I have no clue why the parent shows again in various incarnations under the children, but I'm sure the software engineers that made it - look at it and say "why of course! This is so simple! This is how your data is stored in the system! Why WOUDLN'T you want to see it??" BTW I ran out of space in the screenshot. The "alternative" structure keeps going on and on. For two parts!! Mind you some of those nodes have something unique. In order to start a workflow for releasing a new version or such - I have to pick a SPECIFIC node, and sometimes more than one, in order to kick off the correct workflow. Thanks PLM makers! Great job.

How it got this bad? Some history (from my perspective)

Like many in my generation - my initial experience with CAD in university was with the iconic Solidworks - the successful creation by Jon Hirschtick and team who more recently also created another market changing tool - Onshape by PTC . Kudos deserved. More on the latter - later.

In my first role as mechanical designer in military/defense - I used older versions of PTC Creo: ProEngineer. Wasn't as user-friendly, and took probably two months to get to the same proficiency I had with Solidworks, but the capabilities and overall performance were a step up. That was also the time that PDM was kindda already an obvious necessity (as opposed to using unruly shared drives and manual duplicates). The first PDM tool I was introduced to was Intralink. Probably most of you never heard about it, but it was the best PDM tool you wished you had now. It was dead simple. It was the absolute bare minimum functionality a CAD user ever needs. It had literally 5 commands: Check-in, Check-out, Lock/unlock, New version/revision, Restore. That's it! WYSIWYG. Your CAD structure is the ONLY structure. No "alternative" BOMs, no additional layers of non-CAD data (that user is aware of), nothing to interfere with your natural flow of work.

But alas - that utopia was short lived. It's unclear who killed it, but from what I understood - there was no option to continue using Intralink, since PTC was vested in a new tool called "Windchill". I remember attending with my colleagues to some early demos, where the salesman tried to convince us this was the next big thing. It was "web-based" (-ish, kindda, not exactly), and was pitched as the "everything system" that will not just manage CAD data - but also connect everything and everyone in the organization - from engineering to ERP, inventory, purchasing, accounting, management, everyone. Oh, and it's no longer "PDM". It is "PLM". Product Lifecycle Management. O-kay...

All I remember is I didn't get it. I still don't get it.

Indeed we didn't buy into it right away. We tried to hold on to Intralink. Desperately. But it was clear Intralink was not going to be supported much longer, and after a short failed attempt of in-house software team to bootstrap a makeshift "PDM-esk" system - we gave up and caved to that, um, "PLM thing" that was taking over the industry.

The simplified explanation (speculation) to that odd and seemingly forced anti-user change throughout the industry, is that CAD was a slow market at the time (like most times), and all CAD companies needed new growth avenues - so they all banked on this "PLM" as the everything-system that would multiply the number of users in any hardware company using it. Using CAD? You must want PDM, right? Can't have that. You must have the whole package. You must get onboard the PLM train.

Of course they pitched it as nothing but goodies. Instead of exporting STEP and PDFs you could in "one click" share engineering content with anyone! Well, ok, maybe more than one click. And yeah, not exactly "share" because security and stuff, so more like send to some depository and then go through a multi-step authentication and approval process, and maybe do some manual steps in between, but other than that - super simple! Promise! Also the PDM functionality you wanted - was, em, "expanded" to cover much more "comprehensive workflows". All "digitized". Except for when you really need to get some urgent parts released for fabrication and your boss is out and about, so now you need to find an admin that will circumvent for you the rigid release sign-off workflow. Simple! Simple.

And everyone in hardware industries latched on this bait. Everywhere. Before they even realized - not just number of users of the more-than-PDM system was multiplied overnight, but also the number of Admins needed to babysit those soul draining systems.

Few years went by and I changed roles and joined Tesla, where I had to use Catia for the first time. And Enovia. Learning Catia was almost like learning to walk again. It was an exhausting experience. Unlike the jump from Solidworks to ProE, the jump from ProE/Creo to Catia was much more painful. Other than some anecdotal strengths (yeah, surfacing) - Catia V5 and the "3D Experience" (V6-ish but not really?) were slow, buggy, and often very unintuitive. Ironically - Enovia felt just as bad as Windchill! It's almost as if they were following the same thought process, hmmm... I vividly recall the gimmicky landing page in the "3D Experience" with that silly virtual rotating table that shows some random out-of-proportions exploded view of an assembly you were browsing to, thinking to myself - why on earth would they spend so much time developing something that pointlessly unnecessary? Have they run out of bugs to fix?

From Tesla I moved to a new aerospace startup (Apollo Fusion, later acquired by Astra). One of the benefit of a small startup and an early employee was the freedom to craft my own toolset. Initially we used Solidworks. Oh, good old friendly Solidworks. Warm and fuzzy, and... Buggy?! What happened?? That was not the Solidworks I remembered! The new Solidworks had issues. But worse yet - Dassault was on a quest to coerce its own customers into dropping Solidworks and "strongly considering" Cat... uhm sorry, "3D Experience" instead. They weren't very subtle about their agenda either. The sales rep sent license cost increase warnings every few months. One of these messages really pissed me off, enough to take it to LinkedIn in a now-deleted post, where I complained about this practice from Dassault, while also highlighting how Solidworks isn't worth the pay compared to nearly free tools like Autodesk Fusion 360. I used Topology Optimization feature set as an example for an overpriced license, and the infuriating increased costs every cycle. In a semi-viral post and commentary, I was asked nonchalant by Adam Keating what I thought about OnShape. In my ignorance at the time - I dismissed OnShape as a yet-another pointless Solidworks copycat trying to shake a saturated market with nothing new, with no clue of how different it really was, and who were the icons behind it. Instead of publicly humiliating me for my ignorance, John McEleney was kind enough to reach out and explain to me how uninformed I was, and ultimately convinced me to give it a try - which resulted in us switching to OnShape as our main platform. People who never tried OnShape can't appreciate enough the simplicity and outstanding ease of use of this platform as-is right out-of-the-box. Imagine Google Docs live editing just in CAD + base level version control and smooth going back and forth instantly between versions and design alternatives (branches). Literally in single click. Bam! Utopia all over again. Hallelujah!

In an unexpected plot twist - few months later - PTC made one of its smartest moves ever and acquired OnShape. Oh what a small world.

Shortly before Astra acquisition of Apollo, I joined (the now closed) Apple SPG. Once again - I had to learn new CAD package. NX + Teamcenter. This time I came with low expectations after the Catia experience. Surprisingly it wasn't as bad. But still bad. Just not as bad. But it's bad. Actually it's terrible. Just in different ways. Teamcenter feels like it was stuck in the early 90's. The most uninspiring UI and overloaded dialogs I've ever seen. It's almost as if software engineers designed it for software engineers...

As for NX - both NX and Catia have similar tendencies to bury features and tools REAL DEEP such that you must know what to look for or that it even exists. While Catia shines in surfacing, NX shines in solid modeling and direct manipulations. But when it comes to any form of auto-constraints, auto-select, or default orientation and such - NX is unique in that it's ALWAYS getting the defaults WRONG than user intent. Every. Single. Time. Start a new sketch? The orientation is inverted. Pick an edge? It first picks the one right next to it, or behind it. Extrude? Hole? It will for sure default to the opposite direction than intended. Literally every design task with NX takes 4x longer and 10x more clicks than with any other CAD that I've used before. Maybe it's just me?

Luckily I didn't have to learn yet another completely new CAD package when I joined SpaceX. Serenity now.

At this point - nothing can surprise me. I've seen it all. I've used it all. And I have some things to say.

PLM and the Nanny State

To understand my perspective - you need to understand the broader attitude I have towards authority and hierarchy.

I'm what you might call a Conservative Libertarian. More accurately but less commonly used term: Minarchist. Which plainly means: I believe in highly limited MINIMAL government. Not to be confused with Anarchists, that believe in no government at all, neither law or order. Some of them often wonderfully self-proclaim to be "Libertarians" as well. I don't share such views. There is a fine line. I do believe in maximum freedom, so long as it doesn't infringe on others', natural rights, basic law and order, live and let live, non-aggression, voluntary participation, free market and free will. (First Amendment and Second Amendment absolutist without exceptions; we can debate on those specifically over DMs or a beer, out of scope for this article).

Those beliefs extend beyond the larger, sometimes abstract, political affinity. They apply to engineering and corporate culture as well. An overbearing government - Nanny state or worse - is detrimental to freedom, productivity, livelihood and overall health of society. See how California and New York are doing by any measure that matters. Similarly - an overbearing governance in corporate - is detrimental to creativity, motivation, productivity, and overall health of the company. See [pick any large and sleepy corporate that stopped innovating long ago] and how they do compared to companies that continuously innovate and succeed.

How does that translate to engineering? To PLM? Simple. I consider "PLM" tools as a sort of government package. There was a time, long ago, where that optimal, minimally necessary level of governance was already achieved. It was Intralink and its very limited set of functions as mentioned above. ANYTHING that came later - made productivity worse. Every. Single. PLM tool. Ever.

The "original sin" of PLM was to assume that it's inevitable to have to bring (read: shove) all corporate roles under a single umbrella of interconnected database(s), and assume it will just work. In reality - it makes everyone slower.

Design engineers author most content in CAD. Then there's the base level PDM data. Versions/revisions, release status. There is only a handful of parameters that design engineers care about.

Any attempt to cater to more users outside core engineering - results in overloading BOTH engineering and non-engineering roles with distractions. The further away you move from CAD content - the more "tools" proclaim to come for "help". Requirements, quotes, orders, design reviews, internal discussions... Enter: JIRA (and all it's bottom-feeding copycats). World's most infamous productivity killer that even surpasses PLM. Quite a feat TBH. Here's how JIRA is implemented in most organizations: you need something from someone. You send them an email. They add other relevant recipients to the email. Someone then says "please open a Jira ticket". You open a ticket and it sends everyone from the same email thread - a new email notification. Now you have two email threads on the same topic. Every time someone adds a comment in that ticket - everyone from the original thread get new notification. A discussion starts on the Jira log like a bad Reddit message board, and at some point it becomes too confusing so you all go back to email and start another follow-up discussion, that leads to a meeting, with another email thread of meeting notes, that later someone copies the whole thread and drops the notes in yet another Jira comments that sends new email notification to everyone, and so it goes... Then you spend even more time going back to Jira to close/update/duplicate tickets, and instead of that one original email thread and a meeting - you end up with 20+ emails threads, and hundreds of notification emails that are automatically sent to an email folder of it's own because it becomes unbearable distraction. By the time decisions are finally made - you've lost countless of hours of work on doing something that seems like a scene from Office Space (or Idiocracy).

Why are we even talking about JIRA? Because it emanates from the same flawed way of thinking where PLM came from. It was invented and pushed on the market by the same type of companies, that promise to "streamline" workflows, "organize" data, and help you better "manage" everything. They are not the first and they won't be the last. You'll keep seeing more and more of these popping up in the market every few years.

Once again - PLM, JIRA and the likes - are like the Government. You need to keep the size of both to the bare minimum. Anything they promise to do outside the core functions - make things only worse.

Final thoughts

We in hardware industries, don't need new PLM tools. We need LESS tools. If companies that produce CAD and PDM/PLM tools really care to improve customers experience - consider NOT developing any new tools. Instead - go back to the drawing board and start eliminating unnecessary evils that plague your tools.

And we certainly don't need more JIRAs.

Cheers.