[sugar] Privately-owned data on XOs?

Michael Stone michael
Sat Aug 25 21:07:57 EDT 2007


SJ and I took a rather socratic approach in our previous emails, so, for
the record, if nothing else, I'd like to lay out some of the judgements
underlying our previous questions, after which I will respond directly
to your explanation (which I thank you for providing).


My goals in posing these questions with SJ were simply to:

1) make sure that the alternative of pervasive sharing received serious
consideration and that the design we were pursuing *was an intentional
design*, made with full consciousness of the costs, rather than an
artifact of a hidden assumption to the effect that "because we prize
privacy, it must be important and appropriate for our audience". 


2) to make sure that we understood some of the benefits of the
pervasive-sharing model, including:
  the educational opportunities it makes available that are unavailable
  in a default-private workd, such as a better opportunity to learn to
  make good judgements by studying local history, 

  the increased cost it imposes on anti-social behavior, and more
  generally, the "soft security" it provides by giving everyone who
  wants it access to the information required to find, diagnose, and
  propose solutions to communal and individual problems,

  more powerful positive network effects, similar to those provided by
  other environments designed to universally lower the cost of access to
  information and knowledge (e.g. Google and Wikipedia) that everyone,
  including those who desire to maintain their privacy would benefit from,

  and finally, the fact that the change in perspective simplifies some
  of the trickiest design problems we face, including, for example:

    the problem of providing robust backup services, 

    the problem of designing an effective mesh-wide resource caching and
    lookup scheme, and

    the problem of attempting to provide granular information-flow
    control on an architecture fundamentally designed to control writes
    but not reads and to do so along the boundaries of users and hosts,
    rather than between coordinating groups of processes running as a
    single user on a single host.

Now, with this basis, let me respond to your specific comments:

> The bottom line is we need to not just assume but to demonstrate that
> we are providing an interface that offers real clarity to the children
> as to what is going on. If one knows one is in the public eye, one may
> express oneself differently. I don't think we know enough about how
> effectively we are communicating the various nuances of scoping to be
> able to justify making sharing the default. 

This is a convincing argument for waiting, as is the argument that,
intentionally or not, we are heavily invested in a design that attempts
to make privacy the default and we think we understand the direct costs
and benefits of this design.

> When we can demonstrate that kids understand what it means to be
> connected to other people remotely, I'll feel more comfortable. 

I think I appreciate the sentiment here (that we have to a
responsibility to be conservative in the design we produce) but I have
some questions about what you actually wrote. 

First, are we actually intending to try to measure this (or more
generally, to measure what our design communicates and fails to
communicate)? If so, could you say a few words about what kind of a
measurement you're envisioning us making?

Second, I'm not sure that I know which "kids" are you referring to and
what kind of "understanding" you're looking for. For example, haven't
many anonymous individual kids have been collectively redefining and
understanding what it means to be "connected" for a much longer time than
we have, and with greater overall success thus far? Admittedly the
process is not without risks; people have gotten hurt. So perhaps what
you're saying is that we can't afford to take those risks. Is this what
you meant?

> We already have an interface that makes it much easier to share than
> any other I've ever used.

I think we have an interface that makes it much easier to "share" in the 
synchronous sense we've discussed before, but I'm not convinced that it
does the same for long-term large-scale collaboration. Long-term
large-scale collaboration seems to me to occur most frequently in
environments with lots of world-writable persistent space where the
users of the space are able to shape, to fit their own needs, things
like the structure and contents of the space itself, the means for
advertising in and navigating through the space, and the means for
observing both the contents of the space and its participants.

I don't think we're really attempting, at the moment, to provide such a
space.  Nor are we providing many of the tools that you and I, at least,
enjoy using to manipulate such spaces. At best, I think we might be
providing tools that are useful for independently constructing such
spaces anew. 

Alternately, one could argue that the laptops themselves, irrespective
of what software they contain, are the best available tools for creating
and manipulating spaces like the ones I described. Is this the way in
which you're claiming that our "interface" makes it easier to
collaborate than ever before?

> Has anyone actually seen what the folks at CC have been working on?

I have not.

More information about the Sugar-devel mailing list