The initial GP comment made the point that, at some level, science requires trust. And (in the case of TFA) specifically trust that the person making the observations is actually performing the experiments and recording them correctly -- rather than making them up. You can verify and replicate (and we do quite a bit of that, modulo the fact that resource constraints are a huge problem in science) but without some degree of trust you're in trouble.

But the OP's suggestion was that to fix this specific problem of faked observations, you should separate interpretation and observation. I don't see how that fixes this problem at all. And in my view the first step in solving the problem is to come up with some sense of how serious the problem is: meaning, rather than dwelling on each terrible isolated case and panicking, try to determine what the actual prevalance is and what the overall impacts are. With that information you can make resource-allocation decisions in how to address it. The HN response is much too emotional for anything useful to come of it (except for more anger and confirmation bias.)