We currently have a participation rate of around ~95%. There has been discussion about whether we need to change our parameters (5 sec slot time; 16 slots per epoch) to the same as Ethereum (12sec/32slots)
To have a more informed discussion, I created statistics about the performance of different validator groups. A participation rate of 95% could mean 2 extremes: One would be: 5% of all validators are completely offline and 95% get every attestation, and the other extreme would be: all validators miss 5% of their attestations.
As we want to focus on decentralization a core metric to monitor is whether or not small “from home” validators can keep of with the load. For this exercise, I put all the ~107k validators into different categories based on their deposit addresses.
The 4 largest validators are known: Stakewise (34765), Kleros (19200) former xDAI team (4096) Gnosis (4080)
After those I created the categories:
Large: >1000 validators (7 addresses, 13740 validators)
Medium: >100 (84 addresses, 23869 validators)
Small: >5 (221 addresses, 4840 validators)
Subsidy: 1-4 (489 addresses, 1410 validators)
(note that we are giving out 4 GNO for buying a Dappnode and thus we see a high number of addresses with 4 validators)
Next, after creating those categories I randomly selected 100 validators out of each category and calculated different APY over the last 24h and 7 days. IMO the most reliable number here is the return of the median validator on a 7 day basis - and here it shows that even those “home stakers” (category: subsidy) have very similar results as anyone else.
In contrast - if we look at the performance of the bottom 5% of validators we see much larger differences in performance. E.g. the bottom 5% of the “small” category have simply been offline the last 7 days.
In conclusion: individual stakers are able to get very similar results as professional setups. However, given the much smaller amount at stake, it is more likely that they don’t monitor closely or act immediately once a problem occurs. Easy to setup of monitoring for smaller stakers could certainly help with increasing participation rates. However - those numbers don’t suggest that e.g. 12 sec slot times would make a difference. Rather much higher staking requirements. This, on the other hand would also go against the goal of wide decentralization. To me it seems like something to accept that a small proportion of the validators are offline and luckily Ethereum POS is designed to be able to deal with that.
Stakewise Mean:0.13869097469062502
Stakewise Median:0.1188411484375
Stakewise bottom 5%: 0.11426146435156248
Stakewise bottom 10%: 0.115007107453125
Stakewise bottom 15%: 0.115696503484375
Stakewise top 10%: 0.23998620995312503
Stakewise Mean7:0.140759872253125
Stakewise Median7:0.135917006171875
Stakewise bottom 5%7: 0.11349361091852678
Stakewise bottom 10%7: 0.11550851577455358
Stakewise bottom 15%7: 0.11756190108147321
Stakewise top 10%7: 0.17752351015625004
Kleros Mean:0.11202675493125
Kleros Median:0.1168220596875
Kleros bottom 5%: 0.01970565878125
Kleros bottom 10%: 0.022009726984375
Kleros bottom 15%: 0.0520804350546875
Kleros top 10%: 0.23726092832812504
Kleros Mean7:0.10790820848861608
Kleros Median7:0.1194459232924107
Kleros bottom 5%7: 0.024041066861607147
Kleros bottom 10%7: 0.02813926649330357
Kleros bottom 15%7: 0.05638828498325893
Kleros top 10%7: 0.16344132294642857
xDAI Mean:0.1417829399921875
xDAI Median:0.1191495278125
xDAI bottom 5%: 0.1179815004375
xDAI bottom 10%: 0.118293078125
xDAI bottom 15%: 0.118486926203125
xDAI top 10%: 0.2442728053125
xDAI Mean7:0.1412523280859375
xDAI Median7:0.13557187667410714
xDAI bottom 5%7: 0.11755918623102678
xDAI bottom 10%7: 0.11762783173214286
xDAI bottom 15%7: 0.11769207417633928
xDAI top 10%7: 0.18719027688616077
Gnosis Mean:0.135257103053125
Gnosis Median:0.114410590234375
Gnosis bottom 5%: 0.0866320821640625
Gnosis bottom 10%: 0.092913522859375
Gnosis bottom 15%: 0.09426160359375
Gnosis top 10%: 0.236912474234375
Gnosis Mean7:0.11503684128191964
Gnosis Median7:0.11480255786830357
Gnosis bottom 5%7: 0.07965984238727679
Gnosis bottom 10%7: 0.08748004193526786
Gnosis bottom 15%7: 0.09356228347321428
Gnosis top 10%7: 0.163187812359375
Large Mean:0.1358995337359375
Large Median:0.11866913078125
Large bottom 5%: 0.10752716872656248
Large bottom 10%: 0.108640699890625
Large bottom 15%: 0.1089620858125
Large top 10%: 0.23295552748437506
Large Mean7:0.13253967619553572
Large Median7:0.12732457131696429
Large bottom 5%7: 0.1068380179296875
Large bottom 10%7: 0.10779019888839285
Large bottom 15%7: 0.11391980528683036
Large top 10%7: 0.17529801243973214
Medium Mean:0.130750236125
Medium Median:0.117647547109375
Medium bottom 5%: -0.104806328125
Medium bottom 10%: -0.017576986765625
Medium bottom 15%: 0.097834115234375
Medium top 10%: 0.2538588539218751
Medium Mean7:0.11653063
Medium Median7:0.12990723065848214
Medium bottom 5%7: 0.008008864626116073
Medium bottom 10%7: 0.04237263132589287
Medium bottom 15%7: 0.081213823125
Medium top 10%7: 0.17038547287723216
Small Mean:0.108865388415625
Small Median:0.118129193125
Small bottom 5%: -0.09841067607812501
Small bottom 10%: -0.001996001359374994
Small bottom 15%: 0.065345727578125
Small top 10%: 0.23517640876562504
Small Mean7:0.11046892841339287
Small Median7:0.13209989646205356
Small bottom 5%7: -0.0983099538392857
Small bottom 10%7: 0.02734729796651786
Small bottom 15%7: 0.08690576095647322
Small top 10%7: 0.16708130174107147
Subsidy Mean:0.1169377572734375
Subsidy Median:0.118263199453125
Subsidy bottom 5%: -0.10169049764062502
Subsidy bottom 10%: 0.10790889770312502
Subsidy bottom 15%: 0.11448596501562498
Subsidy top 10%: 0.1306421767343758
Subsidy Mean7:0.11179849204598213
Subsidy Median7:0.12797895521205357
Subsidy bottom 5%7: -0.04375425466406249
Subsidy bottom 10%7: 0.05550842046428572
Subsidy bottom 15%7: 0.1051228788560268
Subsidy top 10%7: 0.15558475090401788
