Why Grade Scanning Always Fails

12-Apr-2007

1. False Positive/True Negative

A good scanning system with only dimensional data (i.e. wane, width, thickness, length) will be 80% right without knowing the grade and 20% would be improved if the grade information were known.

A Grading system gets the grade right 85% of the time and wrong 15% of the time.

So, the grading system will improve 85% of the 20% that a dimensional scanner would screw up.  Lets also guess that the average value improvement is 25%.  That sound good, 0.85 * 0.20 * 0.25 = 4.25% improvement.  Wow, that's $4.25 million a year in a big mill!

But alas, the grading system makes a grade decision on all boards, not just the 20% that needed improvement.  So on the 80% of the boards that the dimensional system got correct, the grading system will give an incorrect grade reading on 15%.  Again lets assume the mistake costs 25%.  That's 0.80 * 0.15 * 0.25 or 3%.  That's MINUS 3% by the way.  So now, the benefit is down to 1.25% or $1.25 million a year.  Maybe it is still worth it, but don’t buy a new boat yet.

Vendors want customers to focus on the benefits (i.e. 4%+ improvement).  It is easy to look at an existing dimensional system, pull out the bad decisions and say our grade scanning system will fix most of these (i.e. 85%).  Never will the vendor pull out a good decision and tell the customer that the system will screw up some (i.e. 15%) of these.

Dimensional scanners have the same False Positive - True Negative (I do not know the correct term) type problem, but these systems have matured to where their accuracy (in dimension decisions) is 98% to 99%, so they only screw up 1% to 2% of the time.

Obviously all my numbers are unsubstantiated.  But then again, all the people selling you a grading system use unsubstantiated numbers.

2. Less graders, more technicians

What do you gain when you trade 2 graders for 2 technicians? Nothing.

3. It’s the lighting.

A machine vision guru once told me, “its always the lighting”.

Mills have been listening to dimensional scanner companies for years whine about ambient light, or sunlight or artificial light, blah, blah, blah.

Well, for grade scanning the problem is magnified. One mill commented that they could grow mushrooms in the grade scanner area.

This reminds me of the disgruntled employee joke. “I am treated like a mushroom. They keep me in the dark, they feed me shit and when I get ripe they can me”

4. Experience counts?

When you hear salesmen brag about their years of experience at grade scanning, you need to translate that into, “We’ve been selling grade scanning failures for a long time”.

Things are getting better. Rumor has it that one mill actually bought a second system from the same vendor. I am certain if we looked into that case we would find that the vendor promised to fix the first one, if the mill would buy the second one.

5. Bait and Switch

One of the best sales gambits is to sell a system that is capable of grade scanning in the future. You get the sale of your standard system and you never deliver on the grade scanning. I tried it once and it worked, but I did feel cheap.

We all know of systems that were sold to do grade scanning and failed, now we are being sold system that do not do grade scanning but will magically succeed in the future.

6. Payback

OK, grade scanning has probably matured to a point that there is a positive payback. As long as you understand that you need the full time grade-tuning technicians and if it is not tuned, it will lose more money than it makes.

Another point to consider is the product life. With a mature dimensional scanner/optimizer, you dislike it in 5 years, you hate it in 8 years and it is declared obsolete in 10 year and you finally replace it in 12 years. An immature grade scanning system will be replaced or upgraded in less than 5 years.

Rod Nelson

“Re-lie-ability. The ability of a salesman to lie to a customer, sell something that does not work, then lie again and sell something else”