At some point you can’t do anything but moving away from home.
I don’t have a above average failure rate but I have problems keeping my home installation reliable. As a home hoster you have to deal with power outages at home, power outages at your internet provider, hardware getting to hot in summer, visitors unplugging the router from the electricity to charge their phones instead (no joke, I had this problem more than once), iOS updates clogging the connection, Netflix 4k streams clogging the connection, the consumer grade routers coming from your internet provider randomly start dropping packets due to whatever reason after running for too long or e.g. under heavy UDP load (I used to restart my router every night with a cronjob to keep my failure rate low).
You have noise while you sleep and you loose rewards in case you allow yourself to go on holiday for a week and something happens while you are gone and there is no one to call (last charging incident for me was when I was away for 5 days and my girlfriend had someone visiting her). In theory this can all be solved and accounted for preemptively. But in practice you have a much easier life just NOT hosting at home. I thought about this more than once, especially now that no one penalizes me for using Hetzner anymore.
If home hosting is a design goal of the network, then the rewards mechanism should make sure that I don’t loose money on running on imperfect infrastructure compared to pay for colocation in one of the big cheap datacenters. I’m not able to do the exact math since I’m too stupid and there are no tools available, but I can see that my node receives slightly below average era points. I’m also lacking the knowledge and tools to identify the root cause but it’s probably my latency since I’m hosting in a remote country on a residential fiber connection.
It seems clear to me that buying a new machine and sending it to Hetzner or Ikoula would be amortized in 1 month. It doesn’t have to be Hetzner, but it would probably end up in Frankfurt, since anything else will be vastly more costly or further from the other nodes so not optimal regarding failure rate. For instance my local DC would charge me $1000 per month for 200 MBit/s and still have high latency to 90% of the other nodes.
In conclusion, I consider my location a feature and accept less payout out of a slightly idealistic motivation, but this proposal scares me and I will see how my rewards will change if it’s implemented and make a decision as others already implied based on the economic incentives presented to me.
Finally some napkin math that I came up with to reason about the problem while typing this response.
Currently my node received 17% less era points than the best performing node as displayed in the explorer. At 1100 xx payout for me during the last 6 days that amounts to 225 xx less rewards than the top performer or 1125 xx less per month. At the community sale price of effectively 80 cents this is $900 per month in missed rewards. This is my incentive to use a centralized datacenter instead of supporting decentralization.
Since the best performing node takes roughly 8% commission, I have to assume that I actually loose something close to 10% of potential income because I’m staking my testnet rewards instead of simply delegating them. Or in other words, my incentive to do nothing and instead use my hardware to play games is an extra $500 per month.
now come and reduce my rewards.