This will just be a quick followup to my previous superblock issues post. A further explanation, an apology and a way forward.
Magnitude Boost
In Gridcoin version 3.5.1.7 business logic was added to superblock verifications to ensure that the superblocks were valid, presumably by avoiding large project dropouts. This meant that if the average magnitude of all the reaserchers in the block was less than 10 the block would be ignored. This worked fine for a very long time, but as the network grew so did the competition. More crunchers meant less magnitude per cruncher which, after 1.5 years, brought the average magnitude down below 10 and superblocks could no longer be created. As a workaround to this every researcher's magnitude was increased by 35%. This is referred to as the magnitude boost. Everybody is getting overpayed, but at least everyone is getting paid paid and the magnitude unit will make sure nobody is getting overpaid (correction by @dutch).
The workaround, or the problem, came at a very inconvenient time. You see, the real fix is to stop checking these thresholds at all and let the network reach a consensus on its own. There are already project count limits in place so if too many projects fall out then no superblock will be created. However, in order to remove those checks we need to release a mandatory upgrade, something which requires a 2 weeks notice to the exchanges. Since we already had an upcoming mandatory in form of the V8 stake engine it didn't make sense to release two mandatories when V8 was "so close".
We failed to properly communicate all of the above which caused a lot of confusion and frustration. I apologize on behalf of all of the devs (I think) for that. I think we can do much better, see Lessons Learned.
The proper fix for the issue has been added to the upcoming Gridcoin release. When the network switches to V8 blocks superblock average limits will no longer have an effect. The boost will be removed along with this and everyone's mag will be back to normal.
Lessons Learned
What we have taken with us from this incident is that we, the developers, need to be much more communicative when something like this happens. We are actively working on issues as they arise but we tend to assume that others will seek us up for information, and that's something that should not be necessary. The next time something like this happens I will, if I remember, try to make sure that either me or someone else quickly updates the community with information as it arrives.
- An initial post describing the current event. "We have noticed that peoples' beacons are disappearing and are currently investigating the issues and possible solutions", or something similar.
- A more thorough explanation. Root causes,discovered workarounds and a solution plan.
- A post-mortem with historical explanations, a detailed description of the problem and its fix, lessons learned (similar to this post), how to prevent this in the future etc.
With that approach it will be much easier for the community members to get a feel for what's going on and it's easier for them to get involved themselves. Note that due to security issues it might not be possible to delve deep into the issue in the "second" post but we'll try to explain as good as we can.