Tierra can now run on a network, in a limited way.
However, there is a lot of work left. Chapter I exposed the goals of the project currently under work, and these have only partially been reached.

Most of the work that was done during the time I spent in ATR dealt with the interface between Tierra-the-application-program and the outside world. This was the easiest part, and the one on which further work should be built (after the questions raised is section III.3 have been taken care of).
But it is far from being enough to get a viable and hopefully dynamic ecosystem running in cyberspace.
The virtual network simulated by Tierra is very primitive. It does not allow a lot of freedom to the creatures, which have very few ways to use it, or even really realize they are in a environment different from the non-network one. For now they only have one new instruction to play with (the "tping" one).

As the Tierra group had a few discussions on this subject, the reader will not be spared a few of the ideas on what could (should?) be done.

The possible needs of the creatures should be carefully assessed, and the adequate instructions added to the new "networked" instruction set. Part of the work will be to decide what should be given to the creatures to do, and what is better left to Tierra.
For example Tierra now uses the tping instruction through a new specialized register in the cell's CPU, which point to the current IP address in the IPMap. But as the IPMap is updated, the address pointed to by this register might change without the creature being aware of it. But should creatures really play around with IP addresses?

One type of instruction that could be useful would be one allowing creatures to communicate. Some of the work now being put in the I/O system between creatures could be transposed to communication between creatures on different nodes. This new mechanism could be kept distinct from the local I/O, or made transparent to the creatures, depending on the choice of the programmer.

Another problem will be to encourage the creatures to actually make the effort to go on the net. Because communications over the Internet will be much slower than Tierra itself, creatures will have to be able to wait for incoming replies to a request. Perhaps a new flag could be added to the CPU structure, signaling the creature in blocked on reading and is waiting for I/O. Creatures will have to chose nodes on networks that are close to them (in access time) for communication to avoid blocking on I/O until they the time they die...

To summarize the previous lines, this report and the associated code only provides a sketchy basis for the real thing.
To all the people that will be working on the next part, good luck, and have fun! I certainly enjoyed myself doing the first part. Well, most of the time anyway.

And there is still the network ancestor left to write...