The document discusses Triple Pattern Fragments (TPF), which is an alternative approach to publishing Linked Data compared to SPARQL endpoints and data dumps. TPF servers are simpler and have lower processing costs than SPARQL endpoints. This allows TPF interfaces to have very high availability for clients. The document analyzes usage statistics of the DBpedia TPF interface which show it has been widely used with high uptime. It advocates for TPF as a way to make it easier and more realistic to build applications on live Linked Data.