Caching trust rather than content

Caching, one of the oldest ideas in computer science, often improves performance and sometimes improves availability [1, 3]. Previous uses of caching have focused on data content. It is the presence of a local copy of data that reduces access latency and masks server or network failures. This position paper puts forth the idea that it can sometimes be useful to merely cache knowledge sufficient to recognize valid data. In other words, we do not have a local copy of a data item, but possess a substitute that allows us to verify the content of that item if it is offered to us by an untrusted source. We refer to this concept as caching trust.Mobile computing is a champion application domain for this concept. Wearable and handheld computers are constantly under pressure to be smaller and lighter. However, the potential volume of data that is accessible to such devices over a wireless network keeps growing. Something has to give. In this case, it is the assumption that all data of potential interest can be hoarded on the mobile client [1, 2, 6]. In other words, such clients have to be prepared to cope with cache misses during normal use. If they are able to cache trust, then any untrusted site in the fixed infrastructure can be used to stage data for servicing cache misses --- one does not have to go back to a distant server, nor does one have to compromise security. The following scenario explores this in more detail.