Also re DNA testing, of course this is a forever discussion. It is true that ‘breeds’ are a human construct, and no ‘breed’ has 100% DNA that has no crossover with any other breed. All breeds branch off from other, older breeds, and there is common DNA found across the breeds.
But as it has evolved to today, the testing by Embark & Wisdom Panel measures the degree of match against many thousands of individuals documented as representing each of many, many breeds, in their respective, independent databases. Both databases are now worldwide and can often segment breeds by geographic region, since the DNA pool overseas may have differences in the same breed in the U.S. and in other countries.
One of the reasons I enjoy the DoggyDNA subreddit is that these DNA sourcing, classification and accuracy issues are discussed by people who have expertise, even some who have worked in the industry and can share their personal experience. How Wisdom Panel frequently lumps Rat Terrier in with Chihuahua, while Embark can separate them.
Along with lengthy explanations of the genes and modifiers behind color and body type. And extra considerations such as dwarfism, which crops up noticeably from time to time.
So definitely there are some random elements, but the identification is getting better as the database pool of accurately identified breeds, by region, expands. And mathematical and scientific methods of identifying DNA to the database pool improves.
Embark in particular made it a mission even before they started public testing to travel as broadly as possible for their database. Some of their breeds have tens of thousands of individuals identified by location to test against.
But it is a fair point that along with guessing being crazy inaccurate at identifying a dog’s breed mix, even testing shows us that breed identification in mixes is not the way to try to control dog behavior.