Login  Register

Re: Croquet vs Other remote desktop sharing applications

Posted by Howard Stearns on Jan 22, 2007; 11:56pm
URL: https://forum.world.st/Croquet-vs-Other-remote-desktop-sharing-applications-tp125505p125508.html


On Jan 22, 2007, at 9:51 AM, David A Smith wrote:

> Hi Kiran,
>
> In some ways, that is exactly what we are after - a replicated  
> simulation environment. The basis of this approach is that  
> collaboration between people is a persistent process, not a  
> discrete event. Croquet allows the user to create a replicated  
> shared environment where there is a guarantee that any objects and  
> simulations in the environment are identical for all the users of  
> this space. The intent is to foster collaboration using this  
> replicated computation. What we want to do is create a broadband  
> communication channel that goes well beyond simply text, voice, or  
> even video - but allows the user to express and share even complex  
> ideas via the simulation.

Continuing this high-level explanation just one level of abstraction  
down, there are several beneficial outcomes from this replicated  
computation concept:

* The programming model (in which the system replicates external  
messages for you and the math gives you the replicated state) is much  
simpler than ad hoc state management by the programmer. Aside from  
productivity, this can have a huge effect on correctness, and  
therefore on project success. The effect is analogous to the benefits  
of compiler-controlled stack management in what were once called high-
level languages, or the benefits of runtime-controlled memory  
management in garbage-collected languages. By contrast, state-
management is the overwhelming majority of  large-project programmer  
activity today, and the cause of the majority of project failures.

* In its current form, the model requires centralized processing of  
messages only to receive, timestamp, and distribute them. This is far  
more efficient in terms of throughput than centralized computation  
systems. Croquet can therefore potentially handle a lot more. Small  
dedicated hardware router boxes are even possible. (Theoretically, it  
may be possible to also be more efficient in terms of latency by  
using a distributed router, but this remains to be determined.)

* The other two benefits makes it practical to include disparate  
models within the system itself, including a model of each user.  I  
believe this provides a tremendous but hard to quantify benefit in  
making applications individually far easier to use successfully,  
allowing applications to be reused and recombined in ways not  
intended by their original authors, and in using them better because  
they are in a more complete social and technical context.

Now, the difference between theory and practice remains to be seen  
here. Also, through extraordinarily careful programming, other  
architectures can duplicate some of these benefits in special  
circumstances. But over time and over multiple projects, it's hard to  
compete with good math.

>
> We have a great interest in integrating existing media into  
> Croquet, but see this as an extension to the primary goal of  
> fostering deep collaboration.

I've been discussing some work with embedded apps, especially  
including VNC. I want to emphasize that I see this as an expedient  
escape to existing stuff that doesn't follow the Croquet model, and  
that such escapes can interfere with maximizing all the above benefits.

>
> David
>
> kiran wrote:
>> Hi,
>>   How different Croquet is from say netmeeting/Remote Desktop  
>> Sharing?   When everything could be shared between machines using  
>> RDS, what additional value does croquet brings in apart from  
>> Modelling and simulation environment. Do we have to restrict  
>> croquet for modelling and simulations and use the existing RDS  
>> solutions to colloborate with other non-croquet  
>> applications..Whats the business value in going for Croquet?
>>
>> Thanks for any inputs...
>> Regards
>> Kiran
>>
>>
>
>