Anyway, to comment on your thoughts about alerts. I think we will increasingly see people switch off alerts altogether. I find that it is rare that the information in an alert is worth more than the time taken to read the alert (let alone the time involved in the context switch associated with the alert). There are two exceptions – an alert relevant to the immediate task at hand – for example an update on a flight departure when I am waiting for that flight, and a “Class A” alert something of immediate personal relevance (eg an accident to someone in the family).

So, rather than an explosion of activity in the status message related tool space, I predict the opposite. I think people will tire of the alerts and start switching them off altogether, or at least start cutting them down to a small number of interesting (to them) sources.

Can’t help but say that for a contrarian view on mashing everything up, do see Penelope Trunk’s post:

http://blog.penelopetrunk.com/2008/06/04/twitter-social-media-and-unmashing-the-mashable/

I think the more fundamental question is to be resolved first in our own (i.e. users’) minds. That is what should drive technological innovation. Anything else sounds like the cart before the horse to me (and more consulting bucks, of course, sorting the mess so may be I should not look that gift horse in the mouth, mixing metaphors, or should that be ‘mashing’ metaphors, a bit…)

You commit yourself to being present somewhere for certain chunks of that timeline (be it a crucial meeting, a party, travel or alone time) and prioritise it in terms of how sacrosanct, immovable it is.

When alternative alerts seek to gain your presence for those timeslots, they might become viewable only if the existing prioritisation level were set low enough. If not, then they’d bounce away, you’d never see them and wouldn’t know you were missing anything.

Additionally, the source of alerts could be rated either via automation (based on how many you’d accepted from that source in the past) and via your personal rating (think family messages, summons from the boss etc) so that the barriers to entry to your timeline that you have set could become more rigid for certain sources of messages and less o for more trusted sources.

Overlay on this a map application, so that any alerts that would be geographically irrelevant given your location for that timeslot would be rejected and I can see that alert overload could be quite easily managed though no doubt I’ve overlooked something.

A Status message is just the tip of the iceberg of deep context. Like social graph, working out the context graph of a person must be possible.

A learning recommendation system that is aware of my the social graph and my context graph will help me cope up with all the status overload.