<p>Having just returned from the IPC Apex trade show, and listening to comments from stakeholders who are charged for volunteering their time to develop standards, one wonders if there is a better way.
</p><p>
Having spent four years of my career in standards development, I know the process well. Groups of engineers, often from competitors, gather round in tables to debate the ins and outs of everything from what an end-product should look like and how it should perform to the placement of commas and the meaning of “shall” versus “should.”
</p><p>
The process can – and does – take years, at great costs in time and travel. And the outcome can sometimes leave the stakeholders wanting.
</p><p>
“In God we trust. All others must bring data.” That was the mantra one enterprising task group chair would remind attendees prior to each meeting.
</p><p>
And yet … in the typical debate, honest opinion – not rigorous fact-checking of the evidence – was the norm. And sadly, the end-product is in practice often ignored by most of the companies of those who worked so hard to deliver it.
</p><p>
I am certainly not questioning the integrity of those involved in the process, many of whom I know and hold in high regard.
</p><p>
Rather, I’m wondering whether, in the era of artificial intelligence, there is a better way to achieve the desired results.
</p><p>
I can’t imagine how much it costs for the defense contractors to fill up, say, the J-STD-001 task group room four or more times a year in their relentless pursuit of another revision. I have to think, however, that the Pentagon could instead command its contractors to produce the design, manufacturing yield, and field data for all its electronics products. It could then stuff all that information into a database, corral a few AI experts to ask the right questions, and produce a truly data-driven electronics soldering specification, all up to date and at much lower cost.
</p><p>
I believe standards should represent the minimum end-product requirements for performance. They should be up to date, reflecting the best practices known in as close to real time as possible. And above all, they should truly represent the data, not the best guesses, of their audience.
</p><p>
There’s more than one way to skin the proverbial cat. This would be one way to get there.
</p>Having just returned from the IPC Apex trade show, and listening to comments from stakeholders who are charged for volunteering their time to develop standards, one wonders if there is a better way.
Having spent four years of my career in standards development, I know the process well. Groups of engineers, often from competitors, gather round in tables to debate the ins and outs of everything from what an end-product should look like and how it should perform to the placement of commas and the meaning of “shall” versus “should.”
The process can – and does – take years, at great costs in time and travel. And the outcome can sometimes leave the stakeholders wanting.
“In God we trust. All others must bring data.” That was the mantra one enterprising task group chair would remind attendees prior to each meeting.
And yet … in the typical debate, honest opinion – not rigorous fact-checking of the evidence – was the norm. And sadly, the end-product is in practice often ignored by most of the companies of those who worked so hard to deliver it.
I am certainly not questioning the integrity of those involved in the process, many of whom I know and hold in high regard.
Rather, I’m wondering whether, in the era of artificial intelligence, there is a better way to achieve the desired results.
I can’t imagine how much it costs for the defense contractors to fill up, say, the J-STD-001 task group room four or more times a year in their relentless pursuit of another revision. I have to think, however, that the Pentagon could instead command its contractors to produce the design, manufacturing yield, and field data for all its electronics products. It could then stuff all that information into a database, corral a few AI experts to ask the right questions, and produce a truly data-driven electronics soldering specification, all up-to-date and at much lower cost.
I believe standards should represent the minimum end-product requirements for performance. They should be current, reflecting the best practices known in as close to real time as possible. They should be process-neutral, so not to impede innovation. And above all, they should truly represent the data, not the best guesses, of their audience.
There’s more than one way to skin the proverbial cat. This would be one way to get there.