I would agree, although I could be convinced otherwise if a useful reason for mapping such underground object was presented. I think to be useful, a lot more information would have to be collected for each node and way used to represent such a network.
(UK centric response) I’ve been mapping and analysing water and wastewater networks from the time I recorded them with pen on paper. We have design drawings, survey records, maintenance records etc and still there are many errors. You can’t just join the dots with straight lines, the dots being manhole covers or other surface features currently visible. They get buried too. Someone has mapped some new network close to where I live. I know it is not as recorded by the network owner and operator as I currently have access to that data (but can’t use it to correct OSM).
A manhole cover might give you access to a simple chamber with an inlet and an outlet. A manhole cover might give you access to a complex chamber of inlets, outlets, overflows, pumps, syphons, weirs, backdrops. Little of this is visible to those who understand it. @aharvey link to what worked in Sydney would not work as is in the UK. Not all sewerage is gravity. Some is pumped and under pressure, some is vacuum pipes. Also what is ‘sewage’? We identify foul water, surface water and combined contents. This impacts on what happens to the network in storm events. A ‘sewage’ definition is not really very useful - similar to highway=path, it might be correct, but how does that help anyone? Plus we want sizes, levels, materials etc. This is to have a useful record.
So, how does recording such networks benefit OSM and the users of OSM? Aren’t there enough above ground features yet to be recorded without introducing what you can’t observe? I’m all for mapping features that can be seen and I have done so. They may help navigation. Recording a water or sewage pumping station as a fact is reasonable, but you can’t assume where it is pumping to with a way linking it to another sewer or water network feature.
Similarly I’d map UK Combined Sewer Overflows as they have a link to other open data sources AND there is clear public interest (and huge understandable ignorance) in their impact.
Obviously if someone wants to map appropriately licensed data of underground networks, then none can stop them, but I’d just hate to see another record that would no doubt introduce new errors and well as duplicate existing ones.
UK Centric mini-rant over. 
Adrian