You don't want people to start questioning your brand (Amazon Best Seller in this case). We all know that World Record Holder doesn't mean much on its own. You have to specify whether it was most pennies stacked on a toe, biggest soap bubble, etc. What TFA points out is pretty much that the title Amazon Best Seller is useless without specifying a category, or possibly not at all.
The deception is him telling people he's a "Best Seller" when really he's "Best Seller in Category." And frankly, Amazon's styling encourages the deception.
Categories are normal in modern bestseller lists. Take a look at NY Times' Bestseller Page:
http://www.nytimes.com/best-sellers-books/overview.html
There is the Fiction and Non-Fiction list, but besides that are dozens and dozens of categories. And each of the authors in each category is a NY Times Bestselling Author.
Now, the NY Times at least has some gatekeeping done by human beings, Amazon has been easy to game for about as long as they've had bestseller lists. It's the same for all their "best selling" categories, some are just easier than others.
By owning one small, niche category with good timing and an easy push, if it drives enough sales, you can basically trade up in your other categories. If you get the right combination and self-sustaining beyond your local allies, you could get traction across the board and build a legitimate best seller in an ever widening space which just furthers the cycle. It's the same as Hacker News.
And it reminds me of another book I read last year.. ;)
Now, with the advent of Amazon, anyone can be a best seller based on very fine grained categories (as shown here) but claim the same accolades based on the title of Best Selling Author. A lot of people accept the title at face value. Over time, this shifts the power from the traditional lists to Amazon since Amazon has most of the power in determining who gets a "best seller" label and at what granularity.
I think this is what the author is really concerned about. In the article he mentions his job consists of working with authors to promote their books and their metric of success is aiming for the NYT best seller's list. If that list loses its value and the cache of being a NYT Best Selling Author decreases, then the author's value add to an individual author is diminished thereby threatening his underlying business model as it stands now.
It's basically the same deal as Amazon, but on a bigger scale. Unfortunately it's not that much harder to game if you have spare cash and/or plenty of friends and/or followers and/or a significant marketing budget.
Not many people understand that sales can peak and dwindle rapidly. Unless you're J.K Rowling or George R.R. Martin, many books typically get a week of marketing effort at most.
So it's not unusual to hit the lists for the first week with - say - 10,000 copies, but only sell another 5,000 - 10,000 copies in the next 12 months.
This may be enough of a reason for a publisher to drop an author. Even after a book gets the coveted NYT best-seller tag, total royalties may still not cover a modest initial advance.
Staying on the lists for a long time is a whole other game, and much more of an accolade than getting on the lists in the first place.
It's not a bug at all, it behooves both parties that this occurs. The seller will sell more books appearing to be reputable, and amazon makes more money because they are selling more product. It's really a buyer beware tale if anything, online retailers are becoming 'used car salesmen' of the 21st century.
How is it a bug in their algorithm? If anything it's a bug in their process. OP found a niche category with so little competition that it was simple to legitimately become the best-selling author by virtue of having the best-selling book (over some time period) in that category?
I agree with him that the title is meaningless, but it's not necessarily inaccurate. What Amazon needs to do - that is, if they care about the 'validity' of the best-seller title - is better vet submissions and purchases. One person purchasing multiple copies should probably only count as one purchase for the sake of this distinction, although that might hurt small retailers that buy from Amazon directly.
I see this as not much different than claiming you're a World Record Holder because you stacked the most pennies on your big toe (it was 21, for the record).