After going through some factors, use cases and friction points, perhaps it would be beneficial to sum up the goals we are going for with any work generation changes. These can then be more effectively used to justify the benefits from any planned updates.
Make receiving as easy as possible
With work generation acting as resistance it is important that changes made avoid disincentivizing receiving of incoming pending blocks in order to keep the ledger cleaner and smaller for those able to prune.
Better work management during saturation
The larger impacts come from having to make work generation adjustments in response to a saturation event. The more flexibility honest actors have to adjust work during this time the easier the network can provide higher quality of service.
Flexible pre-caching options
Remove the current pre-cache advantage spammers have over typical users and services by providing more flexible options to prepare work ahead of time.
Work generation incentive alignment
Remove the barriers preventing those with the incentive to generate work from doing so, most importantly between two transaction participants which are both online such as a customer to merchant payment.
Mind the hardware gaps
Reducing the gap between specialty and commodity hardware work generation performance will help provide additional protection by increasing friction for those with the most powerful hardware.
These are some of the top goals we have been focusing on for work generation updates to the node and we are interested in hearing more feedback on these.
Some will note the absence of a goal to limit pre-computing and instead the mention of optimizing pre-computing for honest actors. There are a couple reasons for that, one of them being complexity of changes needed to limit pre-computing. Many topics discussed in this series can be improved with simpler updates in the short-term, while limiting pre-computing is a much bigger task.
In addition, we largely believe that if the network is properly setup to operate at saturation effectively and all participants have an equal footing for work generation with respect to when they can do it, that the incentive and likelihood of spam being a large issue would be reduced. Whether that is enough to avoid the need for limiting of pre-computing is to be seen. It is clear from the pre-computing poll that there is interest in exploring things further so we will continue to do so.