See this video showing earth pile. Google Earth photos air and ground do not show any at that time. Interesting bunch of composting of city garbage not far away however.
My guess is the earth pile, since I have first hand experience, having been sued due to this effect from nearby fill. Over deep soft stuff the effect goes far sideways.
Something along this line occurred in Albany NY, July 2005. A similar type structure - plate girders, rocker bearings, hammer head pier. At one pier the rockers over extended causing the span to fall off the bearings - it didn't collapse though as it got jammed against the pier cap (talk about being saved by a wing and a prayer). In the forensic report pier flexibility was cited as one of the contributing factors.
This could be interesting legally. In Wisconsin where there is soft stuff underneath a property at the property line, the guy on he next lot can do any filling he wants and is not responsible for what happens next door. Not quite the same for excavations however. However here there was some trespassing.
So far there is no report about pier settling, only tilting. The initial red flag was "cracks formed in ground near the fill". The typical slippage of fill in the circular manner is a sidewise movement of deeper soil, not the surface, in this case likely bending the piling well below the surface. Cracks in the ground at the fill have no relation to possible pile corrosion. My bet is still on. Now, how do we settle bets here???
In watching that video it's obvious there is no R/W fence. Maybe if there was fence along the R/W it might have prevented the dirt from being dumped this close to the piers? Fence is pretty cheap compared to what will have to be done to fix this.
From my experience with an un wanted fill near a building, where soft stuff was 70 feet deep, the settlement effect whent at least 70 feet into the building. For a 150 foot thick soft layer, who knows how far back you have to be to have no effect into adjacent land, especially if there is slippage in addition to compression. Looks like a batch of undisturbed samples will be needed to fully know what happened.
