Lessons Learned From “Whatever Happened To?”

Opinion

 

For some CAD vendors, the tides of time wash away the plans for their future, no matter how well-executed. In 38 years of writing about computer-aided design, I’ve seen numerous companies disappear, perhaps through acquisitions, or else crushed by competitors, and even through pivots by the market. Some were the giants of their times, like Artist Graphics, the nVidia of the 1980s, who threw the biggest parties at CAD conferences (my favorite: the long desert-only tables served on the Queen Mary); today, the Internet has almost no record of the company from Minneapolis that invented the first proprietary graphics chip (see image below; from the collection of Ralph Grabowski).

So, why should we care about companies that no longer exist? Failure does a better job of teaching us lessons than successes do. The outcome of failure helps guide our thinking on the probable trajectories of today’s hot-or-not technology, and makes us able to look past the reassurances by marketing that those newly-launched products are the most important things ever. Until they’re not.

Let’s look at the trajectories taken by several different CAD products.

 

Rejection: Artist Graphics

The first personal computers had miserable graphics, with the original IBM PC offering just 640x480 resolution in a single color. The huge demand from CAD users for huge resolution led to companies like Artist Graphics, Nth Graphics, and Vectrix making fistfuls of money selling high-end boards at an inflation-adjusted equivalent today of $10,000 and more.

Worse, each monitor had to match each graphics board’s sync exactly, making both of them proprietary: changing the graphics board meant spending another couple of thousand on another matching monitor. The game collapsed when NEC released the first multi-synchronization monitor, and then companies like nVidia and ATi (today AMD) emerged to offer low-cost graphics boards and the switch to software-based display-list processing. Artist Graphics spent too much developing their next-generation graphics chip and by 1995 became bankrupt.

Lesson Learned: Being first and holding the best parties doesn’t preserve your place in the market. Stay paranoid by viewing your competitors as better than yourself, and by maintaining a keen awareness of changes happening to your market.

 

Reinvention: Autodesk Nexus PLM

Carl Bass, who was CEO of Autodesk in the years around 2010, considered PLM a joke, initially. Product lifecyle management had been invented by large CAD vendors to get customers to pay for a lifetime of software and services, well beyond the purchase of only drafting programs.

In 2011, Mr Bass changed his tone. Autodesk would have PLM, it would be cheap, run on existing Vault data management software, share data through his online Buzzsaw application, ensure zero customization through hundreds of specialized modules, and be named “Autodesk Nexus 360 for PLM.” As Autodesk saw it, this was in tune with what customers wanted.

Administrator graphically customizing a workflow in Autodesk’s Nexus 360

 

Over time, Autodesk changed the underlying technology, and changed the Nexus 360 name to Autodesk PLM 360, then to Fusion Lifecycle, then to Fusion 360 Manage. Today, Vault is still around, and is sold as both PDM (product development management) and PLM software; Buzzsaw is long gone; and customization is required through programming in Forge (renamed Autodesk Platform Services).

Lesson Learned: What companies think precisely fits the need of customers is not necessarily what customers think they need. When you get it wrong, the product can be corrected, but at the price of customer confusion over product naming and capabilities. And so to this day, Autodesk is a smaller PLM player relative to its size in the CAD market.

 

Misdirection: Solidworks Mechanical Conceptual and Industrial Designer

Dassault was smart when in 1997 it paid $310 million to acquire Solidworks. It became the world’s most popular MCAD program, today generating over a billion dollars a year in sales. For a long time, Dassault left Solidworks alone, but then those millions of users became too tempting a target to ignore. They needed to become users of Dassault’s own software, because it cost more, and more cost = higher revenues.

In 2013, Dassault introduced to Solidworks users the first two programs based on its own software: Solidworks Mechanical Conceptual and Solidworks Industrial Designer. Despite the name, the two had nothing to do with Solidworks, and were based on technology inherently incompatible with Solidworks.

 Sketching a new helmet design with Solidworks Industrial Designer

 

Dassault spent the next decade repeatedly reintroducing variations of its software to the collective yawns of users, and always appending the Solidworks name, such as this year’s Collaborative Designer for Solidworks. How collective is the yawn? So far, 3% of Solidworks users have taken up Dassault’s offerings.

Lesson Learned: Spending an unsuccessful decade pressing your round peg into an acquisition’s square hole means it’s time to concede. Either spin off that balky acquisition, or else make it a fully independent subsidiary.

 

Misapplication: Mobile CAD Apps

Following the launch of iPad in 2011, the future of CAD was to run it on smartphones and tablets, clearly. We would now carry all our drawings with us; it was the end of paper in the field! After all, games and texting worked so well that they were becoming addictive. But not so for CAD.

Many mobile CAD apps appeared in the mid-2010s, from free to expensive. For a time, for instance, IMSI sold its TurboApp for $999, the maximum allowed by Apple store. But after the initial swell of excitement, using a finger to jab at small screens with operating systems unfriendly to the needs of CAD didn’t cut it.

There were problems with limited RAM, the lack of on-board vector processing, and mud-unfriendly hardware. To customers, the value of mobileCAD came close to zero; developers couldn’t sustain themselves on nearly-zero income. Companies dropped out; as the CEO of a competitor told me, "Why should we bother writing a DWG app when Autodesk already did it for us?"

Today, the CAD apps that remain are free, ad-supported, or else bundled with annual subscriptions of desktop software.

Lesson Learned: It takes time to understand how new technology doesn’t necessarily suit your vertical market. Today, when everyone can be a first-mover with all the programming assistance available, it pays to hang back and observe the impact of new tech on your competitors.

 

Misjudged: Generative Design

Probably the most exciting technology in CAD has been generative design, which does what humans cannot: output designs optimized for today’s resource-limited world. This, surely, ought to be in every engineer’s toolbox. Except it’s not.

There are independent firms that offer the software (such as nTopology), and CAD software giants that offer it at extra cost. But when we look about, we don’t see its results in new products. It’s invisible.

One problem is that two technologies have the same name. One generates thousands of alternative designs, whose number overwhelms us ordinary humans. The other optimizes designs by removing unnecessary material, creating models that look ugly to humans, in my opinion. With the ongoing lack of interest, Autodesk, for instance, this year slashed the cost of its generative design extension from $1,280 to $320/year. and PTC is including a version free in Onshape.

Lesson Learned: Customers may not appreciate otherwise-cool technology that adds extra steps and extra cost to product development. A certain level of design inefficiency is tolerated by the market.

 

What Ralph Grabowski Thinks

These examples from history tell us about the future of new technology like AR/VR/XR/MR, AI, NFT, and DeFi. There is an initial exploration phase, in which all kinds of companies work to see how the tech fits: firstly in their marketing plans, and then secondly into their products. Money is spent; money is lost.

The lesson is that new technology, which initially seems to overwhelm the world with its inevitability, always goes in one of two paths: it becomes an along-side technology, or else it disappears. As I write this, corporations are going nuts over AI; eventually, it will fade into the background like the Internet itself, or else fall away like NFTs.

[This article first appeared in Design Engineering magazine, and is reprinted with permission.]

 

Comments

[Simon Weel]

You can start a new topic for this article already. Generative Design will be renamed AI Design - cos we all do AI, right? Just like 25 years ago every software vendor put an "i" in front of their products to tell the people: we do internet! And now every company is putting AI somewhere in their software, err, 'solutions' name. Until AI doesn't prove to much useful at all.

[Ralph Grabowski]

Even worse, some CAD vendors are now claiming, "We've been doing AI all along, we just didn't call it that." Which is funny, it's not like AI is a new thing, say like NFTs.

[Dairobi Paul]

I don't know why I still have this memory. In the late 80's I was a CAD draftsperson using AutoCAD for Wang Australia (remember Wang?) in the department which designed and installed WangNet, a broadband network product which used regular Cable TV coax cable and components. I was in our downtown Sydney office lift discussing with another person the AI potential of lisp. That was over 30 years ago and, yet, AI is new.

[Ralph Grabowski]

LISP was the first programming language meant for working with AI, back in the late 1950s. XLisp was the free version. I remember trying to play around with it before I encountered AutoCAD in 1985.

When Autodesk added XLISP to AutoCAD 2.17f (originally called "Variables and Expressions"), we were a bit surprised, given LISP was meant for AI. But LISP (short for List Processing) was good for handling lists of items, and 2D/3D coordinates and DXF data were lists. It turned out, however, that LISP was clumsy that processing x,y,z data.

David Betz, the developer of XLISP, spent some time complaining that Autodesk had taken his XLISP compiler without formal recognition, and finally Autodesk acknowledged the attribution.

In the end, LISP was a good choice for tossing off a few lines of code inside AutoCAD, because variables did not need to be declared and it was a run-time compiler. For bigger programming projects, it was too slow and clumsy, so ADS (now ARx) was introduced to allow any outside programming language.