What do you mean by "Expensive"? If their goal is to add mod support eventually and they do not have time constraints, why does it matter how long it will take?
Cost in software can equate to many things. Something that is expensive can literally cost money, can take excessive amounts of time, or result in inferior quality. In this example, cost and time constraints are eliminated or reduced, so the depreciation of quality is increased. You may end up waiting a significantly longer time for an inferior solution, because the cost of change over time is exponential, not static or even linear.
A change that costs, say, 10 hours in the design phase could cost 500 hours during development or greater testing. Adding mod support to a big complicated game very late in development could equate to years of work. Furthermore, the cost of a change and the actual work put into a change isn't 1:1, nobody works at 100% efficiency. You're not actually going to spend exactly 500 hours on a 500 hour change. You'll lose some quality. The bigger the change, the bigger the difference, the more quality lost, exponentially. The sooner you start the better.
This is obviously more of a problem when you're designing multi-billion dollar systems for IBM, but I imagine the concept should hold true no matter what software you're building