Re: HML in applications

[Kurt Cagle <cagle@olywa.net>]

> I was thinking that quite a lot of the HML ideas could be application to
appliction service communication.
>
> For example, one service broker could indicate to service requesters the
relative "happiness" of the sites it may call. This could be used for sites
under stress or low n capacity. Even cultural aspects could be potentially
itntroduced, such as "i [the service] am in the US and as it is 9AM, i am
under serious stress".
>
> There are quite a few areas that i oculd see this working, but maybe i'm
getting out of scope !?

"I'm having a bad day, Dave...Would you like to hear a song?"

I think this does touch upon a previous point I brought up. Part of any good
interface design is to present errors that occur in the system to the user
in such a way as to both make sense to that user and provide a means of
diagnosing the problem. I can definitely see an application here:

1) System generates response message
2) Error System Invokes HumanML processor
3) HumanML processor loads user profile, determines skill index in domain
4) Based upon skill index, appropriate diagnostic tools are launched, from a
simple "Your computer has had a boo-boo. I need to restart your application,
okay?" to " Page Fault at 0010:2234 from oxnard.dll. Do you wish to debug or
terminate?"

I've seen some other approaches to this same problem, basically the notion
that the system keeps track of the user's progress (which the user can also
tweak) and can then target the help/error system to provide the right
interfaces. Indeed, the general integration of HumanML into help and system
level resources makes a great deal of sense to me.

-- Kurt