![]() |
Moderated by Stephen Muggleton. |
Erik SandewallOn the Design of Software Individuals |
The
article
mentioned above has been submitted to the Electronic
Transactions on Artificial Intelligence, and the present page
contains the review discussion. Click here for
more
explanations and for the webpage of theauthor, Erik Sandewall.
Overview of interactions
Q1. Anonymous Referee 1 (15.10):
Recommendation: Accept as is. (Details of the review withheld by the request of the reviewer) Q2. Anonymous Referee 2 (15.10):
This manuscript proposes a design philosophy for what the author terms "software individuals". I recommend acceptance of this manuscript after extensive revision. The core of the manuscript is worth disseminating, but it is currently too long and discursive; everything its says could be said in perhaps half the space. The manuscript appears more philosophical than scientific. It would make the paper much easier to understand if concrete examples were used to illustrate the points the author makes. Very strangely, the manuscript does not seem to use the term "agent" anywhere. This is peculiar as it would appear that software individuals are a form of agent. In the related work section, agents should be mentioned and the differences between software individuals made clear. At several places in the manuscript the author tries to distinguish his work from that in the field of genetic programming, this is not always convincing. For example, on the first page it is claimed what distinguishes software individuals from genetic programming is the symbolic nature of the individuals. I am not a worker in GPs, but I believe they would hotly dispute this point, and argue that they often use symbolic descriptions. The paper is generally well written. P2. "We have worked" should be: We have designed P3. Why is small size crucial for self modification? |