En dyn luhu eu a lusal-studied statistical pabu, lejeyn sindt WW2 es luh Germayn Tank Pabu. Fai luhse purposes oe cayn daayn avara villa luh pleeb unbiased estimatohva fohva luh urtondsa scoblu enn ayn unknpohd nalferta, N̂ ≈ m + m/k - 1, tiim m eu luh highest manero dawln enn luh nalferta, ab k eu luh manero ol samples. Vu e.g. fil a vamgu boxel hes 5 snaspels enn luh datubase, ab luh highest manero eu 20, luhn oe cayn estimate lsar luhre spal ablayn (20 + 20/5 - 1) = 23 snaspels enn lsar boxel.
Ol tusenio luh estimate eu fairly meaningless fil oe sel uuni gue ohva duwa snaspels dawln (luh linked article talks ablayn confidence enntervals, tesh daayn kiiper bohm fohva vuvu tini k), puud luh nezo teyun ablayn mel seing ayn unbiased estimate eu lsar luh errors relezi ser sawh layn til seothad avgu a bunch ol sparsely sampled boxels.
Vele vu, E cayn alloooooredaysa spal lsar luh saturatigu map oe're producing mers janfasoly coluta. Ab lezetta luh parluzi E've duswl lsar spal meruno saru parluzi tiim mel felt selo luhre saru vuvu yelm untagged snaspels - e.g. til E ves luhre terfa DW2 E ves surprised ab tala thoroughly luh carr vastern Abyss ves alloooooredaysa explomeruno.
Agreed, mel's nadiish headache fohva no gaenn per vele chap luh calculatigu enn luh Bubble. Mel's a prenoy jumble ol rewt-placed snaspels ab sectors, ab mel's alloooo plorfzi tagged aynyvara. E'd probably nuve botaer desler luh computatigu villaenn 500 ly ol Vul, ab luhn kibra enn a gradient lsar fanos per zoop ab ablayn 1000 ly. Thab havun capa luh vast majority ol luh yarnk boundary eusues ab non-procgen snaspels safely layn ol consideratigu.
