On 18 Jun 2011, at 15:54, Kingsley Idehen wrote:
> On 6/18/11 1:24 PM, Henry Story wrote:
>> On 18 Jun 2011, at 13:20, Kingsley Idehen wrote:
>>
>>> On 6/18/11 12:16 PM, Kingsley Idehen wrote:
>>>> On 6/18/11 8:58 AM, Henry Story wrote:
>>>>> The recent discussions on this list were very much about how to avoid
>>>>> making distinctions unless you have to (Just-In-Time Distinctions?) So
>>>>> why are the above distinctions needed? Particularly with regard to this
>>>>> conversation.
>>>>>
>>>> A root of these conversations lie confusion that results from conflating a
>>>> variety of things. If we separate items into appropriate boxes we stand a
>>>> chance of clarity en route to success.
>>>>
>>>> There are deep unresolved matters that will trigger threads likes these,
>>>> repeatedly. My conflation list is in my last post :-)
>>>>
>>> *At* the root of these conversations lie confusion that results from
>>> conflating a variety of things. If we separate items into appropriate boxes
>>> we stand a chance of clarity en route to success.
>> Every distinction comes at a cost. Say it takes 20 minutes to explain to
>> someone that where they saw As there are in fact A1s, A2s and A3s . Now say
>> you need to explain that to 1 billion people. That is 333 million hours of
>> time taken to explain that distinction.
>
> Come on! Why no earth would I seek that? Is that what you gleaned from my
> comments?
No of course not. I am just developing the thought that distinctions also have
a cost, and that can explain why ontologies tend to have different shapes the
closer you get to the consumer - ie, the larger the crowd of people you need to
teach distinctions. At this point one needs to reduce distinctions - and so
increase fuzzyness - in order to make it more difficult for things to be
misused.
> [snip a big piece of text we know we agree on]
>> Of course if there are 2 people, a teacher and a listener that is then 666
>> million hours taken to explain this at a cost to the economy of 7 billion
>> dollars (if we take the low salary of $10 an hour).
>
> Lost me, I am a little more confident about the inherent intelligence of all
> human beings. The variable that most overlook (IMHO) is attention.
exactly what I was saying: Attention has a cost. And teaching new distinctions
requires time.
> Attention is a critical factor re. perceived human intelligence. This
> fundamental misconception of human intelligence is something programmers have
> become dangerously intoxicated with. This is why programs start to fail when
> end-users become engaged i.e., that hit all the subject matter / domain edge
> cases and to the programmer they are now become super intelligent in totally
> transcendant ways, basically a nightmare that typically leads to solution
> implosion.
>
>> So the distinction would need to generate more value that that to be worth
>> growing. Now of course in a computerised world, the teaching part can be
>> automated, so that perhaps after covering engineering costs the whole cost
>> to the general economy is 4 billion dollars. If the distinction then helps
>> make the interactions between all those users more than 4 billion dollars
>> more efficient, especially if this is distributed around to each individual,
>> then the distinction has a chance of spreading that wide.
>
> Now we're talking! You are describe value of the kind delivered by solutions.
> Yes, delivering useful solutions that leverage new frontiers, insights, or
> tweaks of what already exists == potent education mechanism. Users will be
> engaged and competitors alerted re. opportunity costs.
:-)
> Er.. how about using URL when talking about addresses and data access?
> Speaking about URI in generic sense is problem #1 when speaking outside this
> community. Doing that is lazy, careless, and really unacceptable IMHO. This
> particular tendency just drives people nuts.
Agree. One can use URLs most of the time. The specs are reasonably abstract,
but in context one usually need know no more than a URL, which most developers
already do. Drag and drop is the gesture for those that don't have that
distinction yet.
>> So when people discuss if the distinction between a URI for an object and a
>> URI for a page is worth making, it really depends to whom.
>
> We are introducing a new aspect of the URI abstraction, but assume the
> audience groks the nuances. That saying a URL is a URI solves the problem
> whereas it does the complete opposite. Discarding URI for URL has the same
> effect, and is the biggest headache I see re. communications since it always
> veers down the: a Car is not a Document path.
>
> A URL conveys specific meaning, and has an established sense with Web users
> and developers. Thus, why not build on that as part of the narrative that
> explains the new Web dimension that puts the full URI abstraction to use?
Agree. Think of my writing "URI" as a typo on my part.
>
>> Initially it may not be worth trying to teach such a distinction to a very
>> large crowd.
>
> Fatal mistake, hence the 12 year odyssey. IMHO
I said "may not be worth". :-)
I think some people believe that this is too complex. I, like you am not
absolutely convinced, but well, one could explore how far one can go without
needing to know about the distinction. My guess is that there are limitation -
not ones that break the semantic web - just limitations regarding what can be
said in a language with less distinctions. SKOS is an example of such a
language: less distinctions, means easier use, but also less inferential
ability.
>
> Many critical players grok the distinction, but for reasons unbeknown to me
> we've decided they don't. They just don't understand W3C parlance. I know
> this from serious field testing with various profiles (novice to guru). W3C
> specs are written in a language that many programmers truly do not grasp.
I agree. A lot of programmers know the difference between many different types
of reference and use those all the day.
> Do you know how a typical (non Web or pre. Web) programmer feels when they
> come to realize that Linked Data is all about exploiting de-reference
> (indirection) and address-of operations via hyperlinks? I am talking about
> folks that have developed serious applications solving serious problems pre
> WWW explosion. These folks are the ones that can easily deliver powerful
> Linked Data applications, with ease, if they understand what its about.
> They'll build applications that trump those built by developers discovering
> and learning the art of programming while within the confines of a new slant
> on an old pattern.
I completely agree that this is not so difficult to understand as people make
out. It just requires the right tools and a the right background. Hopefully the
Social Web can bring this everyday feel to the masses.
>> If one can get their behaviour to be in tune with the distinction without
>> them needing to be immediately aware of it, one can save oneself a lot of
>> money. It is a question of knowing who needs to be tought what, and in what
>> order.
>
> In many cases, the teaching == translation by making connections between new
> terminology and established terminology. Being provincial doesn't make ones
> technology established. I laugh each time I am accused of introducing new
> terminology when in fact I am actually using terminology from realms and
> times pre WWW explosion. Or terminology outside of W3C specs and WWW realm.
Well introducing new terminology does have a cost, and so does translation
between terminologies. As the developer of babelfish.altavista.com I am well
aware that translation can be automated, but the cost is still there, and it is
especially high when automated translation tools are not widely available :-)
>
>> Human beings have managed to get very far on the back of mass ignorance of
>> most things. It is only with the developing technical civilisation that mass
>> literacy had to be brought into place at a huge cost to the state, for
>> clearly even greater benefit. The cost of thinking is great, but most people
>> do learn to use their head, as the advantages provided by it are dramatic.
>> Most people don't know how they think though. So they can think without
>> knowing that much about how they do it.
>
> People need to be stimulated. That's all.
Yes, but see above the relation between the cost of distinctions and the value
of those distinctions to an audience.
And remember the point you made that attention is valuable.
I am just suggesting that many differences of opionion as to the complexity of
the semweb is something that can be modelled in terms of the needs for
different types of distinctions by different members of the community. If we
bring in thinking on terms of costs of distinctions we may be able to bring
some more precise thinking to these arguments.
>
> Steve Jobs figured that out, and Apple continues to take profitable advantage
> of that.
Yes, Apple is very good example. They aim to pushing for maximum simplicity
(Sometimes they go too far) in user interface. But nothing is static in that
space. What is complex at one point, becomes evident a few years later. Steve
jobs said a few times that at the beginning of Apple they had to teach people
to use windows and the mouse! In 1995 I still had arguments with people - a
Professor in fact - who said the mouse was too difficult to use! Of course as
soon as Microsoft had deployed Windows 95 and everyone had a mouse, that issue
really disappeared. The value of the mouse was just too obvious.
That is very similar to many of the discussions on simplicity we are having on
this list.
>
>> So when creating an ontology one could try to design it in such a way that
>> users of those relations would not need many distinctions to get going.
>> "Like" is a good example of something that simple. It builds on the ability
>> of humans to work out what the appropriate object of a "like" is. When we
>> get to computers reasoning in a low contextual space such as the web we need
>> tools such as those provided by the semantic web. Of all possible ontologies
>> (all possible distinctions) some are going to be more valuable to a larger
>> crowd. Then there may be ways even there of reducing the distinctions needed
>> to teach such a crowd. Using DocumentObject ontologies with relations that
>> reduce the distinctions needed by a user of the ontology to get it right,
>> might if done right not reduce the inferential ability of the system that
>> much whilst reducing the need to teach many people some distinctions.
>>
>> It may be that the subject worth developing is such a psychosocial economics
>> of ontology development, which takes the cost of distinctions into account.
>>
>
> Ultimately, it boils down to solutions and adaptable narratives. What we can
> do is issue edicts about the evolution of the WWW. There is a nuance laced
> evolution taking place re. the Web. The real Web 2.0 is the Data Space
> dimension, it precedes the Knowledge Space dimension's broader and clearer
> manifestation.
>
> If we just understand that solutions stimulate end-users and that a
> stimulated end-user is an engaged agent, our narratives will be more flexible
> and ultimately more productive.
>
> Nobody is going to win the: best Linked Data or Semantic Web definition
> contest.
> There are some who will score big re. best of class solutions for exploiting
> prowess inherent in InterWeb scale Linked Data and the inevitable InterWeb of
> Semantically Linked Data :-)
Some parts of the semantic web will use solutions that can make very subtle
distinctions, others will try to develop ontologies that require less of them.
The foaf ontology for example gets it right when it only models that foaF:knows
relation, and avoids going into many details of types of relations.
>
>
> Excuse typos etc.. I type fast, kids yelling, I am also in transit with
> family across southern UK :-) Off for another walk in quiet Devon!
Enjoy your weekend!
All the best,
Henry
>
>
> Kingsley
>> Henry
>>
>>> There are deep unresolved matters that will trigger threads likes these,
>>> repeatedly. My conflation list is in my last post :-)
>>>
>>> --
>>>
>>> Regards,
>>>
>>> Kingsley Idehen
>>> President& CEO
>>> OpenLink Software
>>> Web: http://www.openlinksw.com
>>> Weblog: http://www.openlinksw.com/blog/~kidehen
>>> Twitter/Identi.ca: kidehen
>>>
>>>
>>>
>>>
>>>
>>>
>> Social Web Architect
>> http://bblfish.net/
>>
>>
>>
>
>
> --
>
> Regards,
>
> Kingsley Idehen
> President& CEO
> OpenLink Software
> Web: http://www.openlinksw.com
> Weblog: http://www.openlinksw.com/blog/~kidehen
> Twitter/Identi.ca: kidehen
>
>
>
>
>
>
Social Web Architect
http://bblfish.net/
