[EM] Delegable proxy/cascade and killer apps

[Raph Frank]

On Sat, Sep 20, 2008 at 3:34 AM, Michael Allan <mike at zelea.com> wrote:
> Raph Frank wrote:
>> V(N) is the value per user if N users are using the system.
>> P(V) is the number of people who would use the system if it had value V
>> C = cost per user (time, direct cost etc.)
>>
>> V increases with N and P increases with V
>>
>> The critical mass, Nc, is the lowest N where, it is worth it for the
>> N+1 people to use the system
>>
>> P(V(N) - C) > N+1
>>
>> Ideally, this should remain true for all N > Nc
>
> The equation might not be useful for low values of N.  It does not
> capture N invariant values to participation (term Vconstant?), nor
> cross-population variance in V, Vconstant, and C.  At low values of N,
> these variables may strongly influence the growth dynamic.  For
> example:


True, it was just an approx.

P(N) > N+1

is probably more accurate :p, but gives less info.

Maybe, it would work better if it was a sum or some kind.  The value
function is unlikely to be equal for everyone.

> So the network effect V(N) may have no bearing on critical mass (Nc).
> Nc may simply be zero.

True.  In a situation where one person publishes and there are many
readers.  The value to each reader doesn't depend on how many others.

However, that isn't a peer to peer system.

>
> Maybe the key thing is the distribution of values V(0), V(1),
> ... across the population.  How hard is it to find people for whom
> V(0) exceeds corrected cost (C - Vconstant)?  Where are those people
> located in the social space?  Then repeat for V(1), V(2), ...

True.  I think that V(0) for you system is very low as it is a
communication system.

However, maybe V(10) would be high enough so that lots of people would
consider it worth their while.

> It probably gets *easier* as N increases - average V(N) probably
> increases with N.  But that does not imply a critical mass Nc...  not
> unless we factor in my search patience, if it's a finite quantity. :/
>
> (Or maybe I just better stop analyzing, and "go do it"... :)

Right.  OTOH, you still need to ask what is in it for individuals.
Most people will stand back and let others do the work.  The global
benefit is all well and good, but there also needs to be a short term
benefit.

(Hence my comment on the free rider effect).

>> That is true.  Maybe have a link to the "owner's last edit" and a way
>> for the owner to just click 'approve', which would just edit the page
>> without actually changing anything.
>
> Right.  Like "touching" a file in a code build, in order to update its
> mod time.

Yeah.

>
>> Alternatively, maybe it automatically goes to that page, rather than
>> the newest revision.  The history would be split into older -> current
>> -> newer.
>
> Then the newer (unapproved) revision is not the default view for the
> casual reader.  OK.

Yeah, and if you click history, it defaults to that position too.