InnoVal Systems Solutions, Inc. will modify its original filter design so that it does not block the propagation of queries or query responses that are being passed through the computer on their way to other computers on the Gnutella network. The filter will, however, prevent children from originating inappropriate queries and, more importantly, from seeing inappropriate query responses to their own queries. This will be done in a way that will have no adverse effect on the network or any running servents (server/clients).
There remain two problems: One, inappropriate queries are visible in the monitor windows of servent applications. Two, by not blocking propagation, inappropriate queries will reach the servent and, in theory, allow the user to share a file that a parent, implicitly or explicitly, does not want shared. We are exploring two options:
1) Provide two APIs: The first API would test for the filter and determine if it is running in lock down mode. The second API would test any query to determine if an upload is appropriate. This approach would require voluntary cooperation by servent developers. In addition we would provide Java classes for the Java-based servents running with Windows runtime environments. We would produce a public standard to encourage other filter companies to do likewise so that a single call would work with any servent that chooses to incorporate this feature.
2) Do all propagation at the filter level of those queries and query responses that are deemed to be inappropriate and do so in full accordance with Gnutella specification and protocol. This is more foolproof but requires that the filter become a surrogate-servent. Architecturally, this is a poor idea and it introduces a liability for introducing down-stream error. For instance, BearShare, sends messages to other BearShare servents.
We value advice from the user community and from Gnutella developers and are open to other suggestions on these problems.
We are also exploring several ways to benefit the Gnutella community with out technology. The filter, having no propagation blocking, would be beneficial to some users who want to limit visibility of responses beyond what is provided for in some servents. We might strengthen some of the algorithms, particularly as they apply to spam and hijacking (a special, growing and difficult problem). Another option, high on our list, is to give the user the option to hide himself or herself from web-based