A major problem with text based app interaction is learning the language and predictability.
If I've never seen "let's talk it out" as a feature, then I could innocently type this and see this weird thing that's using my data.
Add to this the NLP component, and suddenly there's a whole corpus of things I could say that accidentally trigger some action. Learning this corpus would be akin to learning a language, but without formal instructions (I still don't really know what Siri's full command set is).
I like a lot of the ideas (3rd party tool tip integration, ideas to improve the bus app, and Foursquare direct call outs) but I think a lot of this would be better in app form. The real issue is installing/uninstalling and unclear permissions models - a Wechat style lightweight app install system could alleviate a lot of that concern.
After all, why do I want an uber notification in Messenger, when I need to have the uber app to order the cab in the first place (that kind of tracking complexity and reliability requirements probably shouldn't just be a web view inside Messenger).
with all these fancy 80/20 implementations there's still an inevitable tacit knowledge aquisition phase where you stop giving it what it tells you it wants and start giving it what you know works for getting the information that you want.
it is for this reason that consistency of, and thorough/feature rich design are all that's needed.
treat it like an API. You can add in the fluff later but stop trying to fluff over the rough bits. Fix the rough bits, and you might find you don't even need the fluff.
I see this attitude among the elitist CS majors who can talk to you about how great the CLR is but are completely divorced from reality evidenced by their use of per-byte event-driven serialization routines instead of, you know, just copying the whole buffer contents at once to their processing buffer. The attitude then manifests as 'well it's only a few thousand bytes a second during communications'.
So I guess I need to come up with a way to market my no-nonsense design principles.
Comments
A major problem with text based app interaction is learning the language and predictability.
If I've never seen "let's talk it out" as a feature, then I could innocently type this and see this weird thing that's using my data.
Add to this the NLP component, and suddenly there's a whole corpus of things I could say that accidentally trigger some action. Learning this corpus would be akin to learning a language, but without formal instructions (I still don't really know what Siri's full command set is).
I like a lot of the ideas (3rd party tool tip integration, ideas to improve the bus app, and Foursquare direct call outs) but I think a lot of this would be better in app form. The real issue is installing/uninstalling and unclear permissions models - a Wechat style lightweight app install system could alleviate a lot of that concern.
After all, why do I want an uber notification in Messenger, when I need to have the uber app to order the cab in the first place (that kind of tracking complexity and reliability requirements probably shouldn't just be a web view inside Messenger).
with all these fancy 80/20 implementations there's still an inevitable tacit knowledge aquisition phase where you stop giving it what it tells you it wants and start giving it what you know works for getting the information that you want.
it is for this reason that consistency of, and thorough/feature rich design are all that's needed.
treat it like an API. You can add in the fluff later but stop trying to fluff over the rough bits. Fix the rough bits, and you might find you don't even need the fluff.
I see this attitude among the elitist CS majors who can talk to you about how great the CLR is but are completely divorced from reality evidenced by their use of per-byte event-driven serialization routines instead of, you know, just copying the whole buffer contents at once to their processing buffer. The attitude then manifests as 'well it's only a few thousand bytes a second during communications'.
So I guess I need to come up with a way to market my no-nonsense design principles.