Here's my impression of the 300MB view of the situation, probably mostly wrong.
Someone at IBM decided that there was money to be made in hyperlocal weather forecasting, and sent their corporate raider department to secure the pioneers of that concept, Weather Underground. In late 2015 they succeeded, and put someone in charge of the recently acquired WU who either didn't want the job, or was sorely ill-equipped to accomplish it (although I'm sure there was an impressive degree on the wall). For one reason or another, the programmers that built WU into what it was vanished, and were replaced by others who quickly a) Lost control of the system and, b) Tried to increase profits by cutting features and increasing advertising, undoubtedly at the direction of the aforementioned uninterested or ill-equipped MBA. Unsurprisingly, WU's website popularity went in the crapper, and the people who made the whole hyperlocal concept possible, us, were really pissed.
For whatever reason, in January of this year, IBM decided to make an attempt to salvage the rapidly deteriorating situation and either hired new programmers, or gave the existing programmers an actual budget and a degree of autonomy to improve the site back to usability. The welcome efforts of those programmers are more apparent with each passing week, and the site is much improved from where it was 6 months ago.
Having said that, the system still has many flaws, and new flaws drift in-and-out of the picture as they work. They are obviously still hampered by an inadequate budget, evidenced by the fact that changes appear on the live system long before they should, but at least they HAVE a budget now and are openly communicating with us in the "Let Us Know Your Issues" thread as they work. Although less than ideal, we are apparently their beta testers, and although progress is slower than any of us would like, it sure beats the alternative.