[Sociam-soton] Re: Is email a social machine?

Wed Oct 30 12:46:33 GMT 2013

On 30 Oct 2013, at 11:05, Kieron O'Hara <kmo at ecs.soton.ac.uk> wrote:

> Hmm, not sure about this. Sorry to continue to be boring, but I don't 
> seem to be able to stop.
And then we would miss your pearls of wisdom.
Me sorry too...
========================================================================
The description at the end of your wise message seems to describe classes and instances.
Class defined purely in the sense of the set of members.
There is the class of all (or classes of subsets of) B&D workbenches, and a B&D workbench.
And the class of all (or classes of subsets of) spam operations, and a particular spam operation.
=========================================================================
And possibly going slightly off topic again:
Wendy mentions the idea of a SM “going wrong”.
I think this is an important issue.
There seems to be an idea around that SMs have objectives; which means that they can go wrong.
Certainly the Spammers would not think of the email SM (if it is one) as going wrong.
And the objectives of the creators of email would have been quite nebulous in terms of what would be “going right” and even more nebulous in terms of saying what might be “going wrong”.
It sort of suggests an ownership of SMs by the person or people who create them.
But that begs the question of whether there is even an identifiable community that creates them.
For all I know there may have been people helping to create the email world whose sole objective was to create an environment for spam.
OK, unlikely, but certainly true of anything else you want to think of as a SM.
You can bet the spooks in Egypt etc. were well-involved in using social networks etc to their ends.

Having got this far, I think it is also the implication of a value judgement that worries me about “going wrong”, although that is not a necessity if you define “wrong” as different from the objectives.
But I think it is more helpful to think about complex non-functional requirements (in software engineering lingo) of a SM, and then to discuss which ones are being met, and which are not, what is new and what has changed since the SM’s inception.

I am sure this is all dealt with in the paper or various other papers, but part of the objectives of all this is a social consensus :-)
Best
Hugh
> 
> Is the feedback loop characteristic? Couldn't we have a social machine 
> that left the affordances of the technology unchanged?
> 
> Obviously, any activity on the Web changes the Web in a basic sense by 
> leaving traces and links and altering the amount of information that can 
> be gathered from it. But if (say) a group of people on the Flowers 
> estate used Facebook to coordinate their responses to crime, it would 
> leave the affordances of Facebook unchanged in any serious sense, 
> wouldn't it? But we might still want to call it a social machine, even 
> though the feedback loop is very tenuous.
> 
> Are there any examples of tech/social feedback loops that do not produce 
> social machines? Not entirely sure, but here is one possible example: an 
> al Qaeda group dedicated to blowing up communications lines in 
> Afghanistan (there are such, apparently). There is a set of technical 
> affordances, allowing Web-based and other communications between remote 
> areas and sectors of government. The group blows up the cables, 
> stations, etc, which are so sparse these will seriously disrupt Internet 
> communication (social behaviour changing the technical features). New 
> infrastructure is built with new security features, which then means the 
> group has to alter its tactics but continues to disrupt the Afghan Web 
> (technical features changing social behaviour). But I'd be very 
> reluctant to call the al Qaeda group, still less {the al Qaeda group + 
> the Afghan Web infrastructure}, a social machine except in a very 
> degenerate sense.
> 
> So we can have social machines without feedback loops and feedback loops 
> without social machines.
> 
> This is not to say that the feedback loop isn't important - it is. But I 
> would still argue very strongly that it is goal-directedness that is 
> characteristic. And that email in general isn't a social machine, 
> although individual email systems may well be (maybe in the same way 
> that an individual Black & Decker workbench is a machine, but the 
> totality of B&D workbenches isn't), and email will obviously be part of 
> the structure of very very many social machines. I'd also, incidentally, 
> argue against spam being a social machine, whereas individual spamming 
> operations such as those Dave describes in the citizen science example 
> would be.
> 
> OK, I'll stop now and lie down.
> 
> Kieron
> 
> 
> On 30/10/2013 08:05, Wendy Hall wrote:
>> Below is a thoughtful comment from someone at the dinner last night. It was really cool to be part of a discussion at dinner here and a parallel one on email about email as a social machine. Very spooky
>> 
>> SPAM is often used as an example of when "social machines" go wrong because the designers didn't foresee potential harmful side effects. So I'm veering on email having become a social machine even if it started out as a service
>> 
>> I got confused with the email discussion about halting conditions and outcomes of social machines. Some are purely emergent or evolve to become something they were not "designed" to be and by their very nature may not have a halting condition?
>> 
>> Lots of lovely food for thought
>> 
>> Wendy
>> 
>> 
>> 
>> Sent from my iPhone
>> 
>> Begin forwarded message:
>> 
>> From: Graham Klyne <graham.klyne at zoo.ox.ac.uk<mailto:graham.klyne at zoo.ox.ac.uk>>
>> Date: 30 October 2013 10:00:54 GMT+10:30
>> To: Wendy Hall <wh at ecs.soton.ac.uk<mailto:wh at ecs.soton.ac.uk>>, "David F. Flanders" <dflanders at unimelb.edu.au<mailto:dflanders at unimelb.edu.au>>
>> Cc: David De Roure <david.deroure at oerc.ox.ac.uk<mailto:david.deroure at oerc.ox.ac.uk>>, Kevin Page <kevin.page at oerc.ox.ac.uk<mailto:kevin.page at oerc.ox.ac.uk>>, Dave Crocker <dcrocker at bbiw.net<mailto:dcrocker at bbiw.net>>
>> Subject: Is email a social machine?
>> 
>> Dear Wendy, and all,
>> 
>> (I'm cc'ing Dave Crocker who has a had a hand in email design since before RFC822, and who is also a psychologist by training, so may be well placed to contribute to this debate.)
>> 
>> The question that came up last night about whether or not email is a social machine seems, to me, to be a defining question (among many), because it forces some separation of the characteristics a social machine from the characteristics of the web.
>> 
>> In our discussion, I think you said the defining characteristic of a social machine is the feedback loop between computational system and its social environment:  i.e. technical features giving rise to social behaviours, which in turn result in changes to the technical features.
>> 
>> The more I think of this question in light of the above working definition, the harder I find it is to consider that email is not a social machine.  I understand that some of the original design elements of Internet email were socially rather than technically driven (Dave could expand on that, if needed.)  But also, I can think of numerous examples of peoples behaviour being affected by the nature of email, and consequent changes to the design or deployment of email systems.  I think the most prominent is spam: the nature of email has given rise to spam, which in turn has given rise to technical features such as filters, DKIM, etc.  And the arms race isn't over yet.
>> 
>> I think one of the refutations of of this position mentioned was that email technology is no longer changing in response to social behaviour.  I'm not sure that;'s true, but if it is, I think it gives rise to another question:  do social machines tend to converge to a steady state?  Or if some do and some do not, what are the characteristics that might predict such convergent behaviour?  (e.g. is there some analog of Nyquist.s criterion for linear systems (*) that determines stability or divergence?)
>> 
>> #g
>> --
>> 
>> (*) http://en.wikibooks.org/wiki/Control_Systems/Nyquist_Stability_Criteria#Example:_First_Order_System
>> (My recollecton from undergrad days is that Nyquist's criterion tells us that if there is negative feedback of gain greater than unity for any frequency with with a phase difference of greater than 180% then the resulting system, is unstable.)
>> 
>> 
>> 
>> _______________________________________________
>> Sociam-soton mailing list
>> Sociam-soton at ecs.soton.ac.uk
>> http://mailman.ecs.soton.ac.uk/mailman/listinfo/sociam-soton
> 
> _______________________________________________
> Sociam-soton mailing list
> Sociam-soton at ecs.soton.ac.uk
> http://mailman.ecs.soton.ac.uk/mailman/listinfo/sociam-soton

--
Hugh
023 8061 5652