Those two views can be reconcilled.
Let's start with the "transaction cost" approach - what exactly is being
counted here? From the bartender's point of view, it cost less to
distribute beer to people conveniently placed around a single bar table
than delivering the same amount of beer to these people in their homes.
>From the individual beer drinker's point of view, however, it certainly
cost less to have beer delivered to their home than making arrangements
with fellow drinkers, then physically walking to the bar, then monitoring
the rounds being consumed to ascertain that everyone pays his/her share,
and then walk (or stagger) back home.
It is thus obvious that the argument that a trip to a bar minimises transaction costs holds only if we include the cost incurred by the bartender but not that of the beer drinker in the calculation. Including the latter in the cost/benefit calculus would probably render the bar trip less efficient than drinking alone at home.
That brings us to two fundamental philosophical questions. First, if something is not counted as a cost, does it still constitute a cost? And second, what causes certain costs to be either included in or excluded from being counted in certain cost/benefit calculi?
Economics cannot answer these questions because of its "telescopic" view of rationality i.e. conflating what the economists know from the hindsight with what the human actors actually know while making their decisions. However, cognitive and behavioral sciences (aka sociology) can be of help here. Sociologists have introduced notions such as "social construction" or "definition of the situation" to denote a set of apriori expectations what information is relevant and what can be safely ignored in an interaction. Thanks to that ability we can, for example, have a conversation in a crowded bar - we focus only on what people at "our" table are saying and ignore conversations at the adjacent tables (although we phycially hear them).
In the same vein, the "bar scene" creates a set of socially sanctioned expectations which transaction costs count and which do not: beer drinking and conversation do, while personal effort needed to get to the bar and then go back home (which often may pose considerable problems) do not. Following the same line of argument, managerial ideologies and corporate cultures create expecations which transaction costs count as "real" savings and which can be "safely" ignored as externalities, eg. profit margin vs. employment or environmental protection.
To summarize: economics may explain how information can be processed to achieve optimal cost/benefit balance, whereas sociology can explain which bits of information enter the cost/benefit balancing process and which are excluded from it under different social circumstances.
wojtek