I'm confused...

The claim was:
Since you have a model of the earth where it's a disk, shouldn't the distance between say, the Falkland Islands and South Africa appear different from the distance under the round earth model? This should be calculable based on container ships' routes and so forth.
Which received the response:
Shipping routes are imperfect, and many variables affect the time and direction of a voyage. It's impossible to produce accurate maps using only time and direction. Early maps of East continents were horribly disfigured (when compared to current versions) just by using land travel. Taking compass readings to keep correcting poorly curved paths, accounting for wind speed, or possible currents makes it impossible.
This was then countered with:
But the differences between reality and that proposed by the flat earth are immense.
To which you requested:
Can you provide a single source to back up this claim?
So, to summarise:
The initial claim was that distances between global locations in the southern hemisphere would be calculably inaccurate. The suggestion was made that this discrepancy could be due to difficulties in making accurate measurements due to various forces acting on a ship in transit. This was countered by saying that the discrepancy was far too large to be accounted for by things such as windspeeds and ocean currents. You asked for evidence to support that this discrepancy in distances was as large as had been suggested. I cited dimensions from the Flat Earth FAQ that show large, (in the order of a factor of three, thereabouts), in the distances as calculated by both models of the earth.
For the purposes of clarity, I used the example of an aircraft, rather than a ship, which is less prone to deviation due to environmental factors, but regardless, a difference in distance of approximately 28,000 kilometers is far too large to be accounted for by environmental factors.
This is not even taking in to account that any environmental disturbances would be random, and thus, we would be seeing discrepancies randomly spread across 0 to 28,000 kilometers. To be wrong is one thing, but to be wrong by the same amount every time is statistically unlikely.