RE: [huml] is this a valid prediction?

[Rex Brooks <rexb@starbourne.com>]

This is a really odd-feeling discussion to happen into late in the 
day today, since I basically took today off after the meeting 
arranging for WSRP's 8th Face to Face near-as-dammit week of meetings 
March 1-4. This is year 2 of this effort, and we actually have a spec 
out there, and we are working on 1.1 and planning 2.0 while also 
working on promoting adoption of 1.0. It's odd because we have seen 
quite a bit of attrition due to the fact that web services are not 
"taking off" as the hype two years ago was expecting. That includes 
about half of the 2nd/3rd tier vendors and H-P of the top tier, one 
out of five by my reckoning: IBM, Sun, Oracle, H-P,  MS, with BEA and 
SAP running close behind. MS has never been actively involved in this 
TC, but has had some input now and then.

On the other hand, concepts that were just being being debated back 
then are now part of the landscape, and just as the wind has 
apparently gone out of the sails, the nuts and bolts are now working 
and in the process of being refined. There are lots of bits and 
pieces, like the security services, that are not quite up to speed, 
yet, but that's coming along at an acceptable pace, too, as is ebXML 
Registry and UDDI, so the infrastructure is just now coming into 
being. Personally, I am not at all unhappy that the hype has died 
just as the structure itself is ready to begin widespread deployment. 
I have to agree with Len's analysis, but I think we may be closer to 
reality with this than was expected. As he said, it is hard to tell 
what the proof of the pudding is going to be until you start spooning 
it up.

However, despite MS's truly egregious mediocrity in leaving huge, 
gaping vulnerabilities in its use of ASN, which is about as basic as 
you can get, AND the lagging behind of acceptable security that 
doesn't amount to a moat without a drawbridge, we have VXML 2.0 just 
about ready to go into recommendation status (W3C's equivalent to 
just outright saying, "its a standard,") and THAT is going to make 
the mobile part of the prediction ready to go THIS year, in like a 
couple of months. However that doesn't count for much since it relies 
on the telecom industry for implementation, which is about as bass 
ackwards as we can get. (It's just an aggravation really, having to 
finesse connections for  reliability, security and affordability when 
the Telecoms just really don't have a clue.

WSRP couldn't get them (W3C's VXML WG) to write a portal-specific 
mechanism into 2.0, but it is likely in 2.1 which is already in the 
works, and, regardless, the subdialog construction of the VXML spec 
will work just fine and only needs a minor amount of tinkering to fit 
right into WSRP 1.1, which is probable Q1 or Q2 2005, though we are 
still shooting at Q4 this year. Sooooo, if it takes off, then the 
prediction will be about 200% too long, but I doubt that quite a lot 
just because we won't seea test of major scalability issues in 
practice for a couple of years after the specs are well into the 
early adopter phase--2007 or so. And then it will take about a year 
to adapt. So, my own prediction is 2008 for it to be ready for late 
adopters. XMLP could affect this if it comes out in the next year in 
a form that allows decent performance for asynchronous operations. If 
that happens it might be six months sooner.

It's really fascinating to see the weird timing for all this 
attrition. You can look for platform-independant enterprise 
application integration to be the next big hype, but web services 
will be at the heart of it even if it isn't featured as the key buzz.

Interesting topic. I'm no expert on databases, but I look for them to 
become more focused on Semantic Web inference engine 
interoperability--being able to get the same or similar basic results 
when looking for related topics and resources. Topic Maps seem to 
have become somewhat moribund, which is a shame IMHO.  Knowledge 
representation is another fascinating study. That's why I have been 
preoccupied with RDF and OWL recently.

Thanks for bringing up the subject, James, and for your commentary, Len.

Ciao,
Rex

At 4:25 PM -0600 2/12/04, Bullard, Claude L (Len) wrote:
>It's hard to say.  The push to web services is really happening.  On 
>the other hand,
>the release of the next generation MS system with Longhorn, Indigo 
>etc is won't
>happen for another two or three years.  That's a long time.   If you 
>are asking about
>tightly focused apps such as we build for public safety, that's 
>about two years away
>before we do really noticeable bits although some bits we do today, and we are
>in a late adopter industry.    The bigger event that isn't mentioned 
>in that prediction
>is the convergence of web services and GPS to enable ever more 
>interesting geolocation
>based services based on user preferences.   That is why the 
>distributed identity apps
>are appealing to some, but they scare others.  I suspect that 2010 
>is a good round
>SWAG for this stuff, but I am often surprised as the speed with 
>which these things
>happen.   When I was asked in the eighties what all these 'feedback mediated
>systems' based on global hypermedia systems would do, all I could reply was,
>"Things will speed up and evolution is one of those things.  There 
>are dangers."
>
>We are in a bewildering time of propagandized paranoia and rapid uptake of
>NotReadyForPrimeTime technologies.   I don't even have a cell phone but it
>is because I am a) cheap and b) don't want a digital leash or c) more data.
>
>len
>
>-----Original Message-----
>From: James Landrum [mailto:james.landrum@ndsu.nodak.edu]
>
>Thanks Len,
>Good tight discussion of the issues, and muchly appreciated.
>  I suppose I should have also specified concerns regarding the 
>actualy timeline (by 2010), and am wondering if in the interim since 
>the report was published, (December 2002) if things have changed in 
>the industry that would either (1) shorten the timeframe by one or 
>more years (circa 2008-2009) or sooner,  or (2) extend the timeframe 
>(2011-2012)?
>
>Bullard, Claude L (Len) wrote:
>
>>It is but it is unremarkable as predictions go:
>
>1.  One reason to build a data warehouse is to shield enterprise 
>systems from the Internet.
>Putting mission-critical systems on today's Internet is a profoundly 
>risky idea given DDoS.
>See replication and isolation:  inside the Intranet and on the Internet.
>
>2.  Using a single application may or may not mean a single 
>database.  Usually it doesn't.
>Large system database  builders recognize the need to replicate. 
>Otherwise, locking
>problems are deadly at some n scale (some relational databases are 
>better than others
>at locking).
>
>3.  The generalizations in that description are vast.  "The world of 
>tomorrow will have many
>more people in it."  D'oh.  Yes, the bandwidth is a problem.  D'oh.
>
>4.  What does 'customized access to data in real time' mean?  In 
>other words, a report
>is a report is a report.  A form is a form is a form.   Are either 
>'customized in real time'
>or do programmers customize forms, deliver them, and then the data 
>they 'get' or 'post'
>or 'put' is returned in 'real time' minus network hops, minus 
>processing time on the
>server, minus traffic conditions, and so on?
>
>5.  Data that is not validated before it is stuffed into a database 
>is dangerous data.
>Full stop.  ALL business databases use some form of business rules 
>on the client or the
>server or both validate data.   Sometimes they validate it, then 
>pull it back out
>and send it to another system (think Federal reporting of crime statistics)
>which revalidates it.   Again, d'oh.    So what is the role of 
>synchronization?  I can
>guess but enough.
>
>It's valid in the sense that yeah, that's what we do with databases. 
>How web services
>changes that is a mystery.  Wireless is just another device with 
>possibly server-side  
>XSLT or other means to make the package sent to it light enough to process.
>
>A service-oriented architecture is a big API for exchanging 
>documents be it change
>orders, contracts, or geek love letters.   One can adopt an RPC 
>approach to that,
>but all the Internet knows is bits on the wire and all the WWW adds 
>are ways to
>name those bits on the wire so they can be located, then put, get, 
>posted,. and you
>get the idea.   We used to call these Document architectures before 
>that quit being
>trendy.   As a thought experiment, try to visualize a set of 
>services that an enterprise
>can perform digitally and determine how many of those are not 
>reliant on sending and
>receiving documents.   I want to know the time.  Fine.   Does it 
>send me a clock? Does
>it send me a timestamp?  Does it send me a message instructing the 
>clock on my machine
>to display the time?
>
>The idea behind web services at the base is to expose parts of a 
>database via a set
>of ports with named resources such that anyone can write a client to 
>them.  The hope is
>that a client can then aggregate information from multiple sources, 
>organizations, etc.
>for a specific view.   Think tight fusion from loosely organized 
>sources.  If you think
>in HumanML or ontology terms, there are some semantic issues with doing that.
>
>len
>
>  -----Original Message-----
>From: James Landrum 
>[<mailto:james.landrum@ndsu.nodak.edu>mailto:james.landrum@ndsu.nodak.edu]
>Sent: Thursday, February 12, 2004 3:29 PM
>To: Human Markup
>Subject: [huml] is this a valid prediction?
>
>reference:
>
>Betts, Mitch. 2002. "More Predictions on the Future of 
>Mobile/Wireless Computing." ComputerWorld Magazine, December 16, 
>2002. ComputerWorld, Inc. 
><http://www.computerworld.com/mobiletopics/mobile/story/0,10801,76656p2,00.html>http://www.computerworld.com/mobiletopics/mobile/story/0,10801,76656p2,00.html
>
>
>Quote:
>The marriage of Web services and mobile devices will have powerful 
>implications for enterprise users and IT managers. It will give them 
>customized access to relevant data in real time. For example, remote 
>sales distributors using handheld devices in the field could use a 
>single application that could access inventory records, price lists 
>and customer profitability statements without having to build a huge 
>data warehouse. Also, synchronizing data through a Web services 
>interface allows the application to validate the data before writing 
>it directly to the database, which is critical to maintaining data 
>integrity. However, the limitations of current network connections 
>will prevent the use of Web services for thin mobile applications 
>until after 2010. -- Joe Owen, chief technology officer, XcelleNet 
>Inc., Alpharetta, Ga.
>--
>    
>
>
>--
>From the desk of James [Jim] E. Landrum III
>Archaeology Materials and Database Manager,
>Archaeology Technologies Laboratory (ATL),
>North Dakota State University (NDSU).
>Ph. 701-231-7115  FAX: 701-231-1047
>Email: <mailto:james.landrum@ndsu.nodak.edu>james.landrum@ndsu.nodak.edu
>ATL Web Site: <http://atl.ndsu.edu>http://atl.ndsu.edu
>Digital Archive Network for Anthropology and World Heritage (DANA-WH)
>DANA-WH Web Site: <http://www.dana-wh.net>http://www.dana-wh.net


-- 
Rex Brooks
GeoAddress: 1361-A Addison, Berkeley, CA, 94702 USA, Earth
W3Address: http://www.starbourne.com
Email: rexb@starbourne.com
Tel: 510-849-2309
Fax: By Request