Sorry you feel that way Saiph, but the purpose of this feature wasn't to create a performance hierarchy of literally every available model of GTX 970, but rather to establish whether a difference exists at all - and if so, how much.
We picked out the Black Edition as the winning card because it produced a higher average FPS count than the other cards in every single benchmark, stock and overclocked. If you check out its overclocked values, you'll see we were able to boost the voltage significantly higher than the other cards, and add a fair amount of additional core clock speed to the manufacturer's already boosted defaults.
I hate it when people say "Sorry you feel that way". It's incredibly clear why Saiph was expecting something different from the article, here's why:
"However, the card’s popularity has created an unusual problem for anyone shopping for one – there are now dozens of variations from rival third-party manufacturers to choose from, and the price range for these variations of what is essentially the same card is quite broad – Overclockers UK stocks 29 different GTX 970s, for example, and there’s a £150/$230 difference between the cheapest and most expensive. It begs the question: if all these cards are based on the same reference design, what does the extra money really buy you? Is there a tangible performance gap between different models of GTX 970?
That’s what we’re here to find out."
The way the article is introduced is entirely different from the "purpose of the feature". At the very least it should have been a "cheap", middle of the road, and expensive 970 in the tests, to answer the article's own question: "there’s a £150/$230 difference between the cheapest and most expensive. It begs the question: if all these cards are based on the same reference design, what does the extra money really buy you?"