We have not abandoned work on improving this API. We are still in the planning stages for what we want to release this year, but I am optimistic that it will be a material improvement over what we released last year, especially when it comes to bolstering your own anti-cheat systems.
As for the alt accounts service, we are very aware that this is the biggest issue with what we have released so far. However, I do want the chance to explain why it is what it is right now. We sometimes call it the sibling problem. How should our systems disambiguate the difference between one real person creating two accounts on the same machine versus two siblings sharing the same family computer? To us, this is a big user experience problem that we cannot just waive away, so we try to incorporate other signals that might change our confidence in the accounts being linked or not. With all of that taken into account, our confidence threshold is necessarily very high, resulting in many false negatives.