LightPIR paper says SCC hasn't grown much from March 2019 to January 2021)
Opened this issue · 1 comments
A paper based on this dataset claims something I find strange:
We performed our empirical evaluation on the largest strongly connected component (SCC) in the snapshot and we found that although the network size almost doubles from March 2019 to January 2021, the number of nodes in the largest SCC remains pretty consistent across all snapshots (≈ 2500 nodes).
Is this really the case? As per Table 2 in the paper, on 2019-03-13 the number of nodes in the largest SCC and overall was 2480 / 3480, and on 2021-01-13 it was 2707 / 6376, respectively. Did the size of the largest SCC relative to the whole network really drop from 71% to 42%?
Moreover, based on the snapshot I have personally taken from a c-lightning instance for my research, the largest connected component contained 99.1% of all nodes (December 2021).
Sounds very strange indeed, I'd expect the SCC to be close to 100%, so my guess is that your result is correct. I haven't done any analysis on the datasets beyond some trivial sanity checks, so it could contain some discrepancies, especially on earlier dates.