Gary,

 

Thanks so much for adding the GenCAM part of the story.  Here are a couple of questions concerning the points you made:

 

> Today, with intelligent data such as ODB++ and IPC's GenCAM we are facing the same issues.

> Yes, IPC has again provided a compliance test module to verify the output/input of GenCAM data.

> Yes, some of the output software is being offered for free. Yes, companies are offering tooling

> discounts for intelligent data files. And yes many companies are using either of the two formats.

Could you pass on information as to who is offering free output software?  Also, I would like to know about fabrication shops that offer discounts for use of intelligent tooling.  When you say many companies are using either format, are you talking about fabrication shops?

> Will either format take 100% market share? No! The fact remains that there will be followers of

> both formats, followers of new formats, and then a large group that will do nothing but sit around

> and wait for who knows what to happen.

This is the real tragedy.  Right now, all fabrication shops accept Gerber, as far as I know.  The more different intelligent formats we standardize, the more expensive and complicated the software for the fab shops will become since they will be forced to deal with multiple formats.  That will eat up much or all of the possible savings the smart format was intended to provide.  Consider that a CAM engineer at the fab shop will have to be conversant in each format as well as be familiar with the bugs, tricks and workarounds for each format on his/her importing software.  IMHO, one new format would be a boon to the industry, while several new formats would be a drain on resources.  In that case, we may be better off turning RS-274X into a real standard and create a validation suite for it.  I can't believe I just suggested that in public, but it's better than dealing with a Tower of Babel due to multiple standards.

 

Another issue that I've heard bandied about is the possible merging of ODB++ and GenCAM.  Can't recall where I heard this, but it is an interesting idea.  Is this rumor correct and if so, what is the status of that effort?

 

We can look at the personal computer industry and see examples of how single and multiple standard solutions fared.  In the early 1980's, IBM released the workings of the ISA bus.  Though their information was not totally complete and there was no validation mechanism that I remember, it wasn't too hard to fill in the blanks and it became the de facto standard for about 10 years.  Even after the initial period of heavy use, motherboard and software vendors were compelled to provide backward compatibility for this standard.  Toward the end of the useful life of the ISA bus, it became a bottleneck for increased performance and there was a huge amount of pressure to come up with an alternative.  A number of companies extended the ISA architecture and released the EISA standard.  About the same time, the industry formed the VESA consortium and came up with an architecture that gave better access to the processor local bus.  Also around the same time, IBM released the MicroChannel Architecture because, well, they were IBM.  Despite their technical superiority to ISA, none of these solutions lasted more than a couple of years.

 

Enter the PCI special interest group who took the best of MicroChannel, VESA and EISA and came up with an extensible bus architecture that served the industry well up through the present.  By the time it is superceded, the PCI bus will have been useful for about 10 years.  If you look at the period when ISA, EISA, VESA and MicroChannel coexisted, progress in the industry was stymied by the multiple standards.  Both hardware and software companies were hamstrung and tried to hedge their bets by producing the same product on multiple bus platforms.  In contrast, during both the stable ISA years and the stable PCI years, technical advancements and total sales took off.  Everyone could work efficiently as there was a single hardware platform and the market for every product was larger due to the single standard.  The software situation was not as good because a single company controlled the de facto standard.  This is not inherently bad, but due to their mindset and the lack of broader industry control, they made frequent, undocumented changes to their interfaces and tools that made software development a very expensive endeavor.  Those with limited resources were slowed to a crawl or eliminated.

 

There's a lot we can learn from this.

 

1) It is to our mutual advantage to select a single standard, even if it is not optimal.  Having several similar competing solutions will slow down industry progress.  IMHO, we will do better with a single mediocre standard than several more advanced but competing approaches.

 

2) It is dangerous to have a single company in a position to control the standard.  If their market share is great enough, they could, like Microsoft, participate in industry standards efforts and sign off on the results (i.e. HTML, Java), then go ahead and violate the standard so competing products were not interoperable.  If the standards were not controlled by a couple of players who had large competitive axes to grind, this probably wouldn't have happened.  This is not meant as a criticism of Valor and does not discourage ODB++ from becoming the new standard.  It does have implications as to how the new standard should be managed.

 

Regards,

Seth Goodman
Goodman Associates, LLC
tel 608.833.9933
fax 608.833.9966