CADestiny

Dead Formats Society

At the SME Data Exchange conference, at least two of the CAD vendors were promoting their own "open" data format while simultaneously forgetting to support the STEP format.  It seemed to me that the audience was well aware of the double speak involved in these presentations.

Over at Wired News, Momus published an article called the Dead Formats Society.  Momus was not specifically talking about CAD data formats, but his analogy applies just the same.  He described how the Shinto Ise Shrine  is torn down every twenty years and reconstructed.  The only former items that remain in every shrine are a mirror, a string of jewels, and a sword.  His point is that formats come and go so quickly that we are bound to be losing something important along the way.

In the CAD industry, one of the reasons the vendors do not want to surrender the keys to their data formats is to lock in their customers and continue collecting licensing fees, even after the CAD package is dead.

Momus's title is, of course, a play on the Robin Williams' film Dead Poets Society.  One of the best scenes in that movie has Robin Williams, as a all-boys prep school teacher, showing his class the pictures of the students in the school trophy case.  The students in the pictures were all long since dead, but he pointed out that except for the hair styles and dress they all looked pretty much like the students in his class.  The he said he could hear them whispering "Sieze the day, sieze the day."  We can cease the day too, and encourage our companies to stop using proprietary data formats.

EC: Software is not patentable

Here is an article about the European Commission saying, "last week that computer programs will be excluded from patentability in the upcoming Community Patent legislation, and that the European Patent Office (EPO) will be bound by this law."
EC: Software is not patentable

SME Data Exchange Conference

I am off to the SME Data Exchange conference in Chicago.  I hope to see you there and I will post again when I return.

Pushing a "Pull Model"

Here is a good editorial about the open source business model: Pushing a "Pull Model" by Erik Keller, Wapiti LLC.  As far as i am concerned, it should be called "Eric! Get out of my head!"  What i think this does is complement the interoperability problem with a corporate sales inefficiency problem.

The article reads almost as well as an Isaac Asimov novel.  In the second paragraph,

There is a huge consensus that the enterprise-software industry business model is broken; there is no consensus, however, on how it should be fixed.

It is not exactly the aliens landing and taking control of the world, but for the CAD community this is a big issue.

Eric does a great job of identifying relevant metrics to make his point.  For example,

Most overhead expenses are more associated with the selling of a software license then the deployment of services to install software as well as continuing maintenance payments. Thus when you look at the income statement of software vendors from this perspective, a key measure of corporate efficiency and productivity is the ratio of overhead (S&M plus G&A) to software license sales.

I will do this in an upcoming post.

One thing I think the CAD community already has experience with is Software as a Service.  That is essentially the CAD industry business model.  Eric astutely points out,

What this arrangement also does is to create greater long-term control and revenue management for sellers than does the classic software licensing model. Users must continue to pay buyers as long as the software is used for SaaS. Older models eliminate that requirement. SaaS companies have much more longer-term control over their customers as it is harder to swap out an enterprise SaaS provider than a licensed-based one.

SaaS spells the end of interoperability.  I hope other industries take that lesson from the manufacturing industry.

As for the terminology, "push" and "pull," Eric says,

All of these aforementioned models are predicated on the concept of push-based selling e.g. companies are approached and sold a variety of products and services... Rather than pushing products out to the market, a new generation of companies is beginning to pull customers in via a combination of open source or "free" basic products that can be downloaded over the web as well as "for-charge" enhanced products and support.

Eric's strongest support for giving away the software is that small companies spend 90% of their license fee revenue on sales and marketing.  He says this is because they have to spend 2/3's more on sales and marketing than the big companies.

All these costs are ultimately passed on to the consumer. Since the costs are so high, the seller has to make sure the buyer is trapped into using their software.  Thus,  corporate inefficiencies are a chicken-and-egg problem with interoperability problems.

Amazon's One-Click Patent

You may have heard by now that Amazon's one-click ecommerce patent is up for re-examination.  The story is that Amazon annoyed a Kiwi with slow delivery.  Finding no satisfaction through the usual channels, he set up a website to collect donations to pay the $2520 re-examination fee.  (Google news links)

This was satisfying to me, considering I had just started reviewing Dassault's patents.  Maybe the CAD community would like to do the same thing.

BTW, don't think that Dassaut is just my whipping boy either.  I am going to go through everyone's patents and see if there is merit to the claims too.  Well, maybe it is a little fun to pick on Dassault.

US Patent 5615321

I continued researching Dassault's Catia-related patents.  US Patent 5,615,321, Automatic identification of geometric relationships between elements of a computer-generated drawing, is a little better than their bendy-twisty arrow,  because it took me an extra day to wade through the jargon.  I do not think there is a lot of substance in this patent either.  Today, I rename this patent their "double check" patent.

This patent has one independent claim and seven dependent claims.  I am only concerned with the independent claim in this post.  The full size image of the Mindmap of the independent claim is here.

Claim 1 describes a computerized method for "automatically identifying geometrical constraints (coincident, collinear, parallel, perpendicular, relimited, point-on, concentric, tangent, etc.) between two or more ... elements..." of a CAD model.  The word "automatic" is a little misleading because the identification of constraints happens systematically.  That in itself is not fatal.

It took me a day to figure out what they were doing because of the technical programming jargon in the claim.  All that is going on in this method is that they are checking two lists of model elements (point elements and linear elements) for candidates to satisfy a geometrical constraint and then verifying that that the identified elements mathematically satisfy the constraint within some tolerance.  This does not seem to be anything novel or non-obvious so I think - putting myself in the place of the examiner - that the claim turns on the specific language of claim.  Even so, I think the claim is obvious.

First, the claim covers use of this method in CAD systems.  So any other use of this method in any mathematically analogous implementation is not claimed.  Fine, I hereby claim all non-CAD uses of this method.

Second, both element lists are set up as "n-ary trees."  N-ary is an extension of unary, binary, tertiary, etc. and means that any node in a tree structure can be connected to N other nodes.  Not being a programmer myself, I Googled for N-ary trees and they appear to be common ways of setting up lists for easy searching.  Again, I claim all other non-N-ary methods of constructing lists.  (A couple of the dependent claims have to do with other particular kinds of lists, but I am just making a point.)

Third, if this claim had been for checking one list for candidates then for mathematical rigor, this method is obvious.  I have to think checking two lists is an obvious extension of checking just one list.  If not, then I claim checking n-lists where n not equal to 2.

Fourth, there is no depended concerning the order of the steps.  Since any other ordering of the method is a different method it is not claimed.  Thus, I claim that the mathematical rigor step can occur for the point elements immediately after point element candidates are identified.

What amazes me is how impressive the list of patents covering Catia looked when it was just a bunch of numbers.  Now that I am digging through the documents, it looks like the list is little more than marketing.

Open Source Software: Who Gives And Who Takes

There is a good article at Optimize Magazine by Charles Babcock titled Open Source Software: Who Gives And Who Takes.  In part, the article is about big corporate consumers not giving enough support for freely-licensed versions of software.  I would like to offer a couple of explanations.

First, a corporation is premised on a profit motive whereas an individual is free to be altruistic.  Traditionally, if one corporation writes a piece of code important to its industry but does not give away trade secrets, it will probably assign its rights to another company that specializes in software development.  In the sale, the corporation may agree not to compete with or hinder the assignee for some time.

Second, Sun - a big corporation - released OpenOffice open source.  To justify the business decision, Sun would probably point to the fact that Microsoft already dominates the office suite market so the best way to compete is to give away the OpenOffice suite software and standardize the data formats.  Sun at least gets a foot in the door with customers it would never have had a chance to see with otherwise. 

Third, a big corporation may be locked in to a vendor a number of ways.  The way I carp about most on this blog is through the use of proprietary data formats.  Another way may be through partnership agreements.  A big corporation may be bound contractually not to compete with its development partner.  Thus, it is hard to justify spending money on projects that may compete.  On the other hand, one may question the business judgment of entering such a partnership to begin with.

Finally, despite the constraints, some companies realize it is in their best interest to advocate open source and open data format standards so they do what they can.  of course, they have to do what they do quietly.  Thus, I think that Mr. Babcock somewhat understates corporate involvement with open source software, although I agree with him there should be even more support.

In my case, this blog is entirely my opinion, the corporation I work for has nothing to do with this blog, they do not tell me what to write, and I do not seek permission to post my opinions.  The only restriction I place on this is my agreement to keep their information confidential  Unless and until I see their information officially disclosed to the public I would not speak of anything they do or do not do.

US Patent 6,573,896 3D Arrow

Dassault's patent 6,573,896 is about what I thought it was.  Below is a Mindmanger Mindmap - a fishbone diagram - of how I read the patent claims. 
(Here is the link to a full-size jpeg image.)

The patent is composed of four independent claims and 23 dependent claims.  In this post I am only going to consider the four independent claims and one of the dependent claims.

Claim 1 is a "computer implemented" method of displaying the view from a point along a trajectory in a digital 3D space with a superimposed graphic indicating orientation.  As I said before, this is a flight simulator.  Even if this method had not been patented before, I know the flight simulator game on the original Nintendo did all of this.

Claim 11 is for a "software tool" that displays the graphic image over the view of a digital 3D space from a trajectory that is controlled by an input device.  This is another expression of a flight simulator game.

Claim 18 is for "computer executable code stored on a computer readable medium" that displays the trajectory of the viewpoint as the graphical figure superimposed on a 3D scene, and bends it according to viewpoint rotation speed and stretches it according to viewpoint translation speed.  So Dassault patented a bendy-twisty arrow.

Claim 21 is for the "computer data signal" that represents, essentially, the steps of Claim 18.  The way I read this claim, they are claiming the running code (of claim 18) as a separate invention.  It seems like this is superfluous since claim 18 covers executable code, which presupposes the code runs.

Concerning the dependent claims, I will only note that Claim 8 seems to be meaningless.  There is no "three dimensional graphical figure" mentioned in claim 1.  The three dimensional graphical figure comes up in claim 24.  If anything, this claim should have been dependent on claim 24.

In short, I do not think the substance of this patent would create much trouble for anyone creating a new CAD package.  The only thing Dassault has going for it, as far as this patent is concerned, is the ability to pay for a Borg ship full of lawyers to destroy what refuses to be assimilated.

After I get a handle on what Dassault thinks its intellectual property is with respect to Catia I will update this post with other information about the patent.

Update 5-18-2006

A Friend in the patent bar points out:

Claim 8 does not depend on 24 because the use of "three-dimensional" is just descriptive in 8 and then actually claimed in 24.  And it does sound like a flight simulator with a malleable aircraft.

Patents Covering Catia

Has anyone noticed that 7/8's of "Dassault" is "assault?"  Maybe we should spell the name "D'assault."  Their proprietary data format is holding their users data hostage, so it is fitting.

This is a follow up on my recent post, CAD Software Patents, 1976 to Present.  Out of the 28 patents assigned to Dassault Systemes, Dassault lists 19 of them as covering Catia.  Here is the list:

U.S. Patents number 5,615,321; 5,774,111; 5,821,941; 5,844,566; 6,233,351; 6,292,190; 6,360,357; 6,396,522; 6,459,441; 6,499,040; 6,545,680; 6,573,896; 6,597,382; 6,654,011; 6,654,027; 6,717,597; 6,745,100; 6,762,778; 6,828,974 other patents pending.

<Source: Gehry Technologies>

Normally I would not link to such a list, but the Frank Gehry's architecture is too cool and I probably will not have another chance to link to his web site.  Here is a TenLinks article about his company.

In my next post, I will start analyzing  the 3D arrow patent that caught my attention earlier. 

CAD Company Financials

In my last post,  eNorge 2009 - A Digital Norway I noted that Microsoft's 30% profit margin seems to be driven on the back of individual consumers.  This is because large industrial or government customers can threaten to switch to an open source office package to negotiate for a better licensing fee.  That raised the question of what the profits are among the CAD vendors.  Here is a summary:

                                                               

Company	Market Capitalizaton	Gross Revenue	EBITA	Profit Margin
AutoDesk	$8.90 billion	$1.52 billion	$413 million	21.60%
Unigraphics	Privately Held
Dassault	$6.24 bliion	$1.28 billion	$348 million	17.80%
Cadence	$5.36 billion	$1.36 billion	$367 million	5.14%
Parametric	$1.56 billion	$768 million	$93.5 million	8.10%
Intergraph	$1.17 billion	$579 million	$78.3 million	8.04%

 

I think of a CAD product as a specialized office application that deals with geometry.  It seems logical, therefore, to compare Microsoft's income from its office applications to the CAD industry. According to Microsoft's 2005 SEC 10-K filing, the revenue and operating income arising from the information Worker component (which includes MS Office) was as follows:

                               

	2003	2004	2005
Revenue	$9.1 billion	$10.7 billion	$11. billion
Operating Income	$6.4 billion	$7.4 billion	$7.9 billion
	70%	69%	72%

 

The Information Workers segment includes more than just Office, but I think the numbers show why Microsoft has not taken over the CAD market.  Microsoft's operating income (equivalent to EBIT) is about six times the entire CAD industry's EBITA.

As Evan Yares pointed out in his comment to my post about China and CAD Interoperability, programming a CAD package "is about an order of magnitude harder than any other form of application development..."  So CAD is a lot more work for a lot less money.  (It is engineering, afterall.)

(Aside: It also seems clear why Microsoft's competitors would want to take away part of Microsoft's extremely profitable business by offering open source office applications.  They get to take the wind out of Microsoft's sails and get entrance into previously inaccessible customers in one swoop.  End Aside.)

Clearly, a business has to be concerned with maximizing its own profit.  Reducing upwards of $40 billion in annual interoperability costs (see The Total Cost of Interoperability) has to fall on the CAD users as well as third-party solution supplier.  One way the manufacturing community can reduce its costs is to support the development of a low cost, open-source, open data format CAD product.