At present, data associated with experimental alternatives is just an octet string, plus a hope that experimenters will put some way of uniquely identifying their experiment in the string (preferably at the start). This hope could be made more sure by using a dedicated Experimental type, with elements for a code and a name for the experiment.