Read that article and personally found no practical information in it. The concept of seperating one kind of wear from another exists only in the lab. In the real world its impossible that only one kind of wear happens in a chain and due to the way a chain operates pin/bushing wear is far more pronounced.

Tools are by definition `labor saving devices` and have and always will be partly dependent on operator skill to be most effective. I regularly use a Park Tool chain checker to visually demonstrate to a customer that their chain is finished. When the tool will easily drop into the chain at the 100% wear mark it really doesn`t matter if there`s a 2% error in the tool. If the bike is in for a tune up and its the beginning og the riding season it would be stupid to suggest that they come back in a few more weeks.

But if you want to consider chain tools inaccurate - suggest you have another look at rulers too. Having spent over 15 years in high tech labs - = can tell you for a fact that there isn`t a `ruler` on the market that would be accepted by any lab as an accurate measuring device either. All commercially available `rulers` have manufacturing errors `built in` and materials can also add additional errors depending on material and ambient temperature and humidity.

Its fairly common to use a Vernier instead of a ruler in the shop because its more accurate. However, as far as chains are concerned - the most common practice (since there is a large amount of new chain available in bulk) is to use a length of new chain as a reference. The most common chain seen in the shops here is still 5/6/7/8 speed width and retails for $19.99. A freewheel or cassette is more expensive and any time evaluating wear and any possible need for replacement is usually spent there.

That same logic would also apply to a 9 speed or 10 speed drive chain.