This article on GD&T was written by Rick Hughes, a friend of mine from the ASME standards community. Rick was kind enough to let me publish this article on his behalf. I think you will find it interesting.  

When attempting to implement GD&T, one is constantly barraged with a variety of misconceptions, half-truths and outright errors regarding the concepts of GD&T, the symbology of GD&T and whether or not it is a good idea to use GD&T.

Rather than respond to these objections by entering into a debate (or an argument), it is more productive to "set the record straight" with a clear statement of the facts backed up (as required) with a detailed explanation of the concept or requirement in question.

To this end, this article contains a collection of some of the most commonly encountered objections/misconceptions with a short response to each

I call these "THE SEVEN SILLY THINGS THAT EVERYONE ALWAYS TELLS YOU ABOUT TOLERANCES THAT JUST AREN'T TRUE". If that seems too much of a mouthful, how about the acronym "TSSTTEATYATTJAT"?

True, but so what?  ALL dimensions are perfect (absolute). The important thing is that when we use basic dimensions the part is not required to be perfect.

Dimensions (BASIC or otherwise) are absolute. They establish a target value. Tolerances (Geometric or otherwise) specify how close to the target an actual part (or feature) must come.

Showing a dimension as BASIC simply means that the default ("Title Block") tolerance doesn't apply. One must look elsewhere to determine the tolerance on the characteristic for which the BASIC dimension determines the target.

This is a corollary to "Silly Thing" #1 and is rooted in the belief that making things perfect is expensive.

The fact is that making things perfect is impossible and therefore FREE (what's it cost to do the impossible?).

GD&T is a technique for allowing the maximum amount of tolerance while guaranteeing function. It NEVER allows less tolerance than conventional methods and usually allows more.

How can allowing MORE  tolerance increase costs?

GD&T makes sense in any case where one wishes to provide complete, unambiguous, three-dimensional definition of part geometry and to allow all of the possible tolerance while guaranteeing function. These goals are as applicable to one-of-a-kind items as they are to large runs.

This is the "Those Other Dummies" argument. In using it, the speaker allows as how "we" understand alright (because we're so smart), but "Those Other Dummies" won't or don't, and because they won't or don't, "we" shouldn't use GD&T.

While it is true that not all people understand tolerancing (or, for that matter, any number of other technical subjects) as well as "we" might wish, it is also true that many people in many functions in many organizations understand it quite well. In those cases where ignorance actually exists, the answer is education, not capitulation.

If you assume that your audience is illiterate, why are you sending them written messages? It is self-defeating to assume incompetence on the part of "The Other Guy”. 'We" must assume that the people who receive our drawings are capable of reading them, or why prepare drawings at all?

This kind of argument is fairly common and is advanced against almost any change. Its' basis is the belief that the existing method has been shown to work and we should therefore cling to it.

This argument addresses the wrong issue. The question isn't "Does the existing method work?", but rather "Is there a better way?".

GD&T is clearly a better way:

The person who makes this statement is really saying "I don't understand this stuff and I am unwilling to make the necessary effort to learn it!".

Any skill requires some effort to acquire. GD&T is significantly less complicated than many skills which most people already have (reading, speaking, writing, walking while chewing gum, etc.)

All that is required to obtain a mastery of GD&T is a little time and effort and an interest in the subject.

One usually hears this chant from someone who has recently attended a GD&T seminar and heard someone extolling the virtues of the MMC modifier as a facilitator of functional gaging. The truth is that each of the three material condition modifiers (MMC, LMC and RFS) has a specific purpose:

The purpose of the MMC modifier is to make the amount of a geometric tolerance dependent on actual feature size in such a manner as to provide a MMC virtual boundary (inner for hole, outer for shaft) which doesn't change as feature size changes. This modifier is used for clearance (fit) applications.

The purpose of the LMC modifier is to make the amount of geometric
tolerance dependent on actual feature size in such a manner as to provide a LMC virtual boundary (outer for hole , inner for shaft) which doesn’t change as feature size changes. This modifier is primarily used in applications where wall thickness control is the matter of primary concern.

The purpose of the RFS modifier is to make geometric tolerance independent of actual feature size. This modifier is used for centering applications.

Each modifier should be used in its appropriate applications; none is inherently "superior" to the others.

One hears this “Silly Thing” mainly from tolerance teachers or other “gurus”.

It was established back in “Silly Thing #2” that it is impossible to produce perfect parts, yet here we are, saying that a part must be perfect.  This is a mis-statement of Rule #1 of the Y14.5 standard which states that a feature may not violate a “Boundary of Perfect Form at MMC”.

One hears things like “How can you check a .002 tolerance from a surface which varies .010?”  The answer is, obviously, that you can’t.

The fact is that dimensions and tolerances come from DATUMS, not from part features. The DATUMS are imaginary and perfect.  The actual part features with which they are associated will vary within whatever tolerances control them.

In short, you don’t check a tolerance with respect to a part surface, you check it with respect to a DATUM. Therefore, variations in the datum feature (part surface) don’t affect measurements made from the DATUM.

DATUMS are chosen based on part function;  some part features will be perpendicular to the primary datum, some will not. Required orientation depends on function and, therefore, varies from part to part and from feature to feature within a part.

The number of possible Silly Statements regarding tolerances is finite, but LARGE!  These statements are usually rooted in ignorance and/or resistance to change. The cure for either is the same; EDUCATION. Makers of silly statements need to be informed of the facts and convinced that GD&T is a better, more powerful, tool than conventional tolerancing.

Don't let "Silly Things” stand in the way of rational tolerancing!