Radegunde · @petraclaudia
11 followers · 157 posts · Server rheinneckar.social
Łukasz Dębowski · @ldebowsk
14 followers · 3 posts · Server mathstodon.xyzSome hashtags of my interests:
#entropy #language #Zipf #ergodic #nonergodic #mutual_information #information_theory #repetition #stochastic_process #excess_entropy #kolmogorov_complexity #algorithmic_randomness #large_language_model #neural_network #power_law
#power_law #neural_network #large_language_model #algorithmic_randomness #kolmogorov_complexity #excess_entropy #stochastic_process #repetition #information_theory #mutual_information #nonergodic #ergodic #zipf #language #entropy
hobs · @hobs
408 followers · 762 posts · Server mstdn.social@jan this feels right from a theoretical perspective. You have to cap the throughput/load of the most popular nodes in a #mesh network or #distributed system or #federated system. #Zipf's law demands it. The resource duplication of a federated system creates exponential inefficiencies that must be capped in order to achieve sustainable #scaling and #fairness.
Fig6: https://morse.colorado.edu/~timxb/5520/ho/MeshCap2003.pdf
#mechanismdesign #ActivityPub #Fediverse #fairness #scaling #zipf #federated #distributed #mesh
hobs · @hobs
408 followers · 763 posts · Server mstdn.social
@conradhackett @manlius #Zipf's law in action. This will happen for the #ranked list of the #counts of anything in any naturally evolving system (English word usage, city population, animal counts). This is why there is such a pull towards #centralization for #SocialNetworks. The answer is more servers. More #servers means more large servers which means more #decentralization. Can't fight #SocialEvolutionaryDynamics. Harness it!
#Mastodon #socialevolutionarydynamics #decentralization #servers #socialnetworks #centralization #counts #ranked #zipf
Verwechslungsgefährte · @dichotomiker
86 followers · 55 posts · Server dresden.network