Two smaller issues are raised in the previous panel. One is that the files you send need to be plausible. Do you generally send pictures of your family to your business associates? Maybe yes, but if not, sending them just announces the likelihood of a subchannel. The prior discussion of techniques for other file types might be useful in strategizing plausible files for normal transmission. The second issue was mentioned earlier: If your subchannel encoding involves altering non-predictable data, can an attacker gain access to that same data in other non-identical files? For example, suppose you have a strategy for altering information in transmitted flat-file records. Good enough, so far. But can an attacker gain access to individual records by other means, or at other times? Perhaps you have sent an intersecting record set (either with or without a subchannel), or want to later on. If the alterations are inconsistent in individual records, this provides a clue to a subchannel (obviously, production data changes occasionally, but within some limits).