Ingrid is a research project at NTT Software Labs to design and build a suite of software tools that provide fully scalable, fully distributed web searching/discovery. The basic idea is that sites index their web documents locally, and are then automatically joined into a global distributed search infrastructure. A search on that infrastructure from any location will find all (or nearly all) matching web documents and related search engines. At the workshop, I'd like to say what we're doing with Ingrid, give some information about the infrastructure we will have built at that point (hopefully it'll be something), and discuss what "open" protocols we use for Ingrid, including what extensions we need to the SOIF format (I'm assuming at this point that we will move to the SOIF syntax, though we have a number of new types.) Ingrid uses two separate protocols. One is the Ingrid Access Protocol (IAP), which is a TCP-based thing (fairly similar to HTTP) used to submit items (in SOIF form) to the Ingrid infrastructure, and to search the Ingrid infrastructure. People can use this to build applications on top of Ingrid. The other is the Ingrid Control Protocol (ICP), which is a UDP-based thing used to create, maintain, and search the Ingrid infracture itself.