First, "I did X, then Y increased, then I did X again right before Z, which clearly increases Y, happened, and Y increased again!" is not empirical evidence, and many of Marco's conclusions were very dependent on (I think unreasonable) assumptions that were based on his gut feelings.
More importantly, the argument of "he'd lose customers if he removed the free app" is about long-term, future effects. I perceive that Matt is trying to say that, say 1 year from now, the paid Instapaper app will have fewer customers if there is no free app in that interval than it would have if the free app had existed in that interval. Of course, this is fundamentally impossible to test, and is based on gut feelings, but it was not disproven by any "empirical evidence" presented by Marco. But it's no more unreasonable for any of us to have a gut feeling in this respect. Personally, I believe a lot of Marco's rationale, and do think that he's better off now without a free app, but he certainly did not prove this conclusion.
More importantly, the argument of "he'd lose customers if he removed the free app" is about long-term, future effects. I perceive that Matt is trying to say that, say 1 year from now, the paid Instapaper app will have fewer customers if there is no free app in that interval than it would have if the free app had existed in that interval. Of course, this is fundamentally impossible to test, and is based on gut feelings, but it was not disproven by any "empirical evidence" presented by Marco. But it's no more unreasonable for any of us to have a gut feeling in this respect. Personally, I believe a lot of Marco's rationale, and do think that he's better off now without a free app, but he certainly did not prove this conclusion.