Wednesday, May 21, 2008

Temporal discontinuity

A comment Eve just made on a call prompted this

Consider some magazine at which I've subscribed. The mag only needs my address at the start of every month, just in time for sending out that month's edition.

Their demand curve (I'm sure there is a proper economics term) for my address looks like

If the magazine relies on me actually visiting their site (to view online content etc) as the mechanism by which it can get the most recent address, then they are at the mercy of my erratic visit schedule for the update operation.

If I happen to move in mid February

then I don't get the March edition.


Eve said...

I have tried calling this "orthogonality" but Bob Blakley insists that it doesn't meet the technical definition, preferring "temporal loose coupling". I'm not a mathematician, so what do I know, but I think they're conceptually at right angles!

The ideal architecture would allow for completely separate timing when individuals (a) log in and use a service for its normal interactive (online) offerings and (b) set/change information about themselves, and when services (c) utilize the freshest version of the information in providing its offerings either online or offline.

Paul Madsen said...

I guess technically the two schedules do have a non-zero correlation factor, but the key point is that the correlation factor is not '1' (which some systems presume).

~swl said...

While it's unlikely to be adopted, it seems like the relationship between subscriber and publisher would be more balanced if, during registration, the subscriber provided the address of an "address service". The publisher (or their distributor) would then pull the subscriber's address from that service before sending out the magazine. If you move, you just update your address with the service, and all your subscriptions are updated.

One example would be giving the publisher an i-name, which they could then dereference to a an address service, which would lead to a page containing an hCard. In this scenario you could also change i-name broker and address service without notifying the publisher of every magazine you subscribe to.

Paul Madsen said...

swl, why unlikely?

isn't this the Holy Grail?


~swl said...

While this is the holy grail, I'm not convinced that publishers would be ready to move to this model any time soon.

Paul Madsen said...

Scotty, I claim no insight into the publishing biz beyond reading popular science, but there are pressures (e.g. financial, legislative, etc) that may push them there.

if they see 'holding data' as integral to their business, then yes it will be a push