please ignore typos and stream-of-semi-consciouness… will tidy up this evening…

Mike Masnick: If you have the killer app, why aren’t I dead yet
killer app is a marketing term that does no-one any good. If the technological advance is intrinsically useful, then the killer app will emerge. Hype overshadows users.

“For a successful technology, reality must take precedence over public relations, for nature cannot be fooled.” – Richard Feynman

Brewster Kahle: Preservation through replication
factors that influence preservation: Hardware, programer error, format obsolesence, institutional drift, law/goverment.

Created full mirror of library of alexandria, egypt. 100terabytes! look at the history of libraries – libraries are burned by goverments! people don’t want the past around. make copies in multiple places. Physical installation to squirt back out into space in alexandria.

web has grown up outside of the arhivist’ ecology – the wayback is an attempt to bring it back into the fold.

copyright – archve of scanned documents OCR’d for seachability. . right holders can premeptively sue. cease and desist. put it up anyway…. wasn;t that big a deal… seek forgiveness not permission again! was inexpensive to do, and so was doable as a non-rpofit… therefore was seen as asenstive apporach.. .seen as heling not hurting therefore doesn;t get attacked so much.

interllectual property persevers – creaive commons – “national parks for information” – preseving and serving the information films on wayback has started an explosion of films that reuse te archive footage… cheap doumenetaries, film students, crazy videos… opening up content for recombination and greater creativity.

Cory has much more succinct desc. of Brewster’s talk

Meg Hourihan: user-centred web services for the real world
newsblogger -> pyra/moreover.com: first experience of web servics. what possibilites are offrered by the recomibination of a number of data soruces and applciations around a task that affect reall poeple in thre real world
meg’s defn of web services
machine to machine communciation
ways to hook into systems
open standard
cross- platform interaoperbale and ubiquitous (promise)
“faceless”

what are they trndy? web is the app platform of choice, more devices touch te web, standards are emerging, so is cooperation not necessaruily tech centred advacnements, maybe user/task centred adavcnes
rudeuce costs, expanad/extend business models

risk involved though – why do it? are you cutting you own throat? risk to self-sufficiency of your system – what happens if the web-services you are recombing disappear… or the companies you have outsourced to disappear – small pieces loosely joined replaced by small pieces tightly bound by byzantine service level agreements???

why use web services

to get access to one of a kind services
to get access to data that change rapidly
to take advantage of something yourlocal computer cannot handle… what does that mean??

user taks are joined up on the web, but websites cannot predict those paths. webservices

utopia would mean the joins wouldn’t show.

but who would do this? who would provide this? “infomediary” idea again in another form.

combine multiple sources of data in new and interesting ways. topic or task based products

rather than company based. build off some of the good ideas of appliation-service provision.

so away from the utopia… if it;s useful there’s risk… (esp. outsourcing)
bweare of “proof-of-concept” web services
reliabilty / quality assurance… all these things are jury-rigged… and left behind…

more midwives than nannys… or even absentee parents… one-time assembley
more mundane issues: speed and secuirty – speed of dfifernt connections to diff servcies…

design / page layout implicatiopns.

my question – these are still
a) business-centred drivers for creation
b) or user-drivern, expect the users are super-users who can hack the web services

together… even with SOAP etc… it’s still a specicilist/web-dev/hobbyist level of skill to create anything useful. how can the ad-hoc recombination of content/data/info/apps/services be done by a mainstream, real , normal person? Meg answered that there may be another level of mediator of web-services… maybe like a “moreover” for web services to create ad-hoc bundles.

however – the real-world exmaple that meg used was one of travel frustration – where she wnated to recombine te infomation REACTIVELY as events unfolded. One thing we have learned is that it is HARD for companies to model our experiences to our satsisfaction, and what’s more it’s hard us OURSELVES to truly, proactively model our experiences, our needs, we way we do things. Great things about the web and the current wave of web-services is the ad-hoc, REACTIVE, recombinatin of servcies… how take this mainstrream.. what about the microasfot
work on “inductive UI” could that be a possible methodolgy for providing mainstream recombinatory, reactive webservices to real people – will try and et discussion on this at the BOF this evening.

John Ko: Emerging User interfaces
zoomworld/raskin. -> cincro zanvas web.

today’s UI challenges
data scalability
human to human interaction
information flow – one to few
p2p doesn;t cut it scalability and security (refered to centalisation as a way to avoid chaos – but emergence from chaos is theme???)

need for web-servcies that offer session based interaction to foster collaboration and human to human communcation around content
sacalbel, real-time instead of transactional data services – unifed: creation sharing and collaboration (wikis??? IM???)

going beyond HTML to rend this
lack of sacale content deliv ery
dynamic updates difficult – it’s about packages of documents… transactional centric

new appraoch
information decomposition – exploit for edge-accerlation of web content… finer grain of contetn being sent around than is being describeed by page-centric mechanism of http/html.

spatial interfaces? what question are they the answer to?? they almost killed those kids at the end of Jurrasic Park (“hey i know this, it;s UNIX system!”)

inferred from the fact that we are spatial animals. are we? or are we storytelling, pattern recgonising animals.instantenous, dynamic? – but that is a factor of the plug-in and the way it communicated. not the spatial nature of the display of infomation. infinate zoomable space? unique way of delivering to PDA? yes – delivering information, but not delivering understanding. why is it an infinite usauble space that he has bullet-pointed lists in slides that he is delivering in serial order? it;s because it’s a pattern we’re used to.

fine-grain, dynamic creation. object-oriented. wysiwyg. not wizard/transactional based.

terabytes of information are being generated by digital camera users everyday. what is being done with it? what if you could bring that all into a shared space and have instant intraction with others around that. people do this in effect with file-sahiring, but may be uncomratbale with the concept of sharing without control.

asked question about why the zooming was important – answer was about real estate limits, but also realting to “messy desk syndrome” (cf. gladwell, social life of paper”) allowing peope to build meaning in ways they want.

question from floor referred to jef raskin’s hospital information system – easy to understand in a zoomable UI casue it is geographcially based.

rick rashid – microsoft

moore’s law –“what have you done for me latelY”
graphics in games have changed – not gameplay. is this as much of advance as we think?
the underlying concepts that nderly UNix in 1970’s are not that diff. that they are in 2002
a time-traveler from the 70’s could get by…

“time is now to really rethink fundamental realionships people have with operating systems”

how can we leverage moores law
new forms of input
handwriting
gesture
speech
reading, annotation, note taking – harness power of doodling

new ways to analyse info
-assisted understanding – language recognition (grammar tools in word etc)
– general purpose understanding – summarising automatically
-Mindnet laguage knowledge base – words (nodes) connected by realtionships
mine onformation by looking for actual answer to the given question
allow users to ask questions naturally – information retrieval vs docuement retrieval
“ASkMSR” http://www.microsoft.com/usa/presentations/Breese_MSResearchUpdate_FINAL.ppt
good when there is a lot of information to be cross-corelated
we have so much data in the network so it can be mined usefully fr answers.

modeling user-behaviour using powerful tech
understanding priorities of the user in their context.
there has been a complete switch from historically when computers time was more valuable than users. work by MSresearch called “priorities” – eric horvitz
look at communciation histories, sender, recipient realtionships (from company org charts, and/or from more informal means?)
model I/O to user using this context
http://research.microsoft.com/~horvitz/attend.htm

be aware of office context – “office awareness” use microphone or cam to understand when you’re busy, on the phone, talking to someone face to face etc.

changing realtionship to computer, changing computers relationshoip to it’s context

human scale storage – i can think seriously about keeping everything i ever have or do on my personal storage.

user-centric computing model
task centric vs program centric
user selecs task not application
task creates a context for data access code excutuion and visualisation and I/O in general.
capatalise on natural user contexts

memory – what hav i seen
what have i done in the past
task
people
who am i working with wat are they doing, how are they changing things i care about.
data
what is related to this picec of information

Susan Dumais – “stuff i’ve seen”
http://research.microsoft.com/~sdumais/

*query vs hierarchy*
traditional operating system dominated by “location” and hierachy based concepts
we’ve created a dewey decimla system for data – hierarchies

peope more natural think in terms of association

continuous execution – event driven vs expicit command – cf. Lifestreams??? gelertner
security vs personal security
digital rights management vs centralisd trust
no distinction any more between local and distributed

ui: cognition, perSONALISATION, emotion
Lili cheng http://research.microsoft.com/adapt/

user interface -> visualisation modules-> queries and filters -> mapping, clustering sorting -> data collectors checking changies in system -> back round again in a continuous flow

much more suited to distributed systems.

old OS = determinsitc
new OS = probabalistic

some peope think visually ,some liguitscall, some proceederally – acoomadte them all… model and respond to user behaviour and ways of undertasning things

make systems that adapt and PAY ATTENTION.