After reading the OP, I concluded that there might be an issue in the study. It seems unlikely for these students from the London School of Economics to arrive at such a conclusion unless they lack a proper understanding of how mining pools operate. As a result, I spent/wasted half an hour delving into the research paper to uncover the nuances, and here it is.
This figure documents the concentration capacity of miners based on Coinbase rewards that miners receive from pools. Each month, we sort active miners
by the amount of Coinbase rewards they receive and calculate the percentage of total mining capacity controlled by different quantiles of the miner distribution.
Obviously, I do not doubt that their figures are accurate, and this conclusion
The top 10% of miners control 90% and just 0.1% (about 50 miners) control close to 50% of mining capacity
Could have been guessed by anyone who has been mining long enough, but where did they go wrong?
These individuals seem to be entrenched in the early era of Bitcoin mining. In today's world, Coinbase transactions hold little significance in such research. To simplify, the vast majority of mining pools allocate Coinbase transactions to their own addresses and subsequently distribute rewards from there. Moreover, many mining pools operate on a Pay-Per-Share (PPS) model, meaning they pay out rewards before actually receiving them (or sometimes after). This results in payments originating from a different set of coins than those visible in the coinbase transactions.
Given that almost 90% of all blocks are solved by approximately only 10 pools, it may seem as though these 10 miners control 90% of the blocks. Mining pools typically don't attempt to conceal their rewards, otherwise, they could assign a new address for each coinbase transaction. If this were the case, the entire interpretation of the study would shift drastically. Therefore, while the presented figures may be accurate, the interpretation is fundamentally flawed.
Reading the paper further, I found another critical error in their approach which is linking the origin of exchanges to the location or nationality of miners. Their assumption, based on the use of a Chinese-owned exchange implying the miner is Chinese, or the use of a US or EU exchange indicating nationality, is flawed. Common sense dictates that this inference is far from accurate. Unless exchanges reveal miners' information to these researchers, which is highly unlikely, their methodology only exposes the identity of mining pools rather than individual miners.
In essence, this research may suffice for academic purposes in obtaining a master's or bachelor's degree, using it for anything else -- is a mistake.