There's some nuance. When we talk about points like this, we're talking about very sparsely populated parts of the Earth like Western Australia, Siberia, northern Canada, some of the remote parts of the Amazon, etc. Especially for the points on the coasts and in the extreme north and south latitudes, this is really important. Why?

Because when you do global temperature modeling, you divide up the planet into squares in order for the computer to model them, something like 5 km^2 blocks, but I've seen some that get more ambitious than that. In order to start the model, they have to give the computer a starting temperature value for all of those blocks. But if they don't have temperature recordings from weather stations in each of those 5 km^2 blocks, they have to "approximate." They take the one remote weather station closest and simply fill in the data for the surrounding blocks. So, if you manipulate those sites specifically and make them falsely hot, the impact in the models is greatly exaggerated. Take a site out in the desert of Western Australia and make it a degree hotter and now half the continent plus all of the surrounding water looks hotter.

This is even worse on the oceans. As you certainly know, 70% of the world's surface is ocean. Our measurements are taken by ships who do recordings as they sail along common trade routes. Naturally, there are some huge limitations to the quality of that data, the most obvious of which is that there are huge parts of the ocean where ships just don't go all that often. The same process happens. Some scientist takes the closest available reading and uses it to fill in data for huge chunks of the ocean. They violate the most basic rule in science which is to use mathematical calculations at a far higher level of precision than their data allow.