Thread by @VitalikButerin

This is my approximate view of what the next ~5-10 years of eth2 and beyond might look like.

Roadmap below reflects my own views, others (including future versions of me) may have different perspectives! Details may of course change as we discover new info or new tech.

In general, over the last ~2 years there has been a solid shift from "blue sky" research, trying to understand what is possible, to concrete research and development, trying to optimize specific primitives that we know are implementable and implement them.

We know much more about the capabilities and limits of proof of stake, security models of sharding, etc etc. Thanks to the hard work of many, moon math like ZK-SNARKs is looking increasingly down-to-earth.

(PS: the moon is made of Havarti, not Gouda; we checked)

Additionally, over the last ~2 years there has been an increasing move toward maintaining compatibility and ensuring a smooth transition for applications, as well as making eth1.x and eth2 fit together as part of a coherent vision.

From here, research is continuing, though the bulk of the challenges is now increasingly around development, and development's share of the pie will only continue to grow over time.

In addition, there is less reason than ever to fear ever-growing complexity; the "quadratic sharding as end of history" approach ensures there will not need to be a further upgrade as large as the eth1 -> eth2 one; rather, from eth2 onwards things will be much more incremental.

And many of the changes are actually in the direction of *reducing* complexity.

Look forward to seeing great work continue to happen and for all of this to be put into action!