I always used to read Jeff Masters blog on WU.
Then in the summer of 2012 he posted in his blog about how the site had out grown them and they were selling out to TWC. He promised how nothing would change and the new backing of TWC would make things even better.
I knew right then it was the beginning of the end. The original WU was successful because it was run by weather weenies like us. They gave a dam about the product. When corporations take over all that goes by the way side.
They weren't in it for the money. They were in it because of their passion.
I don't know if there will ever be another site like WU but their death started in 2012. What we see today is the continuing slow death of a once great organization.
My personal opinion, it the sale to TWC was done largely because the revenue sources to fund WU were drying up v.s. the Weather Company's revenue sources.
WU was using Ad revenue and sale of custom weather products to newspapers (and others), but that couldn't quite cover the costs of infrastructure and staff sustainably given the competition by Intellecast, Accuweather and TWC who had similar revenue sources.
WU's offer of 'paid subscriptions' gift to PWS data submitters cut Ad revenues by removing ads from weatherunderground for logged-in PWS oners (cutting back some potential revenue). Likewise, the commercial sales of custom forecast pages to newspapers was declining as newspapers themselves declined in physical-print subscribers. Those, and an increasing cost of the aging infrastructure behind WU likely meant that a 'white knight' was needed to inject funding to continue .. At first, TWC was viewed as that white-knight. TWC, like any corporation acquiring a pre-built functionality, wanted to streamline and port the WU functions to their existing Cloud-based infrastructure. The problem was likely that the WU architecture wasn't either easily ported (due to older non/cloud technologies), nor deeply understood by TWC's developers (which I imagine have split into front-end/back-end/data assimilation/API generation groups, as contrasted with the small, integrated team that developed WU in the first place). This possible scenario would have made any over-arching plan to preserve functions while porting extremely difficult (as we've seen in fact).
Note that the above is only my guess.. not based on any inside-knowledge.