Wednesday, June 06, 2012

#iel12 Wednesday Afternoon Keynote

Blogging and video on an unstable surface + away from a power outlet = awesome.

Hopefully - my battery holds.....

BTW - higher quality videos will be available later and I will provide the link once they are posted.

The Ustream is just me, a cheap webcam, and a really old netbook.
Perpetual beta (for me) often equals amateur video.

IN this case - I forgot to click the Record button, so the recorded session will be missing the first 15 minutes.  (Serves me right for checking my work email.  Sorry Prof. Siemens) And I missed the last couple questions because my battery almost died.
-------------------------
Wednesday Afternoon Keynote - George Siemens
Analytics: sensemaking, prediction and performance

Used to think in the "educational technology" space until 2005
- Bulk of research actually more specific

How do individuals sensemake and way-find to complex info systems
How do we interact with fragmented info sources?

Context
- August 5, 1949 - Mann Gulch disaster
  + Expert created "escape fire"
  +Social / technical system not optimized and not working well

- Dec 2-3 1984 - Bhopal disaster
  + System maintenance reduced. Create system failure.
  + Still social / technical
  + Communication breakdown
  + Technology more prominent

- April 14, 1994 - Black Hawk Incident
  + Massive failure of technical system
  + Some social strands element - but failure of technical

We are dealing with systems where we are looking at sensemaking as a systemic entity rather than individual action

Sensemaking - a motivated continuous effort to understand connections...in order to anticipate their trajectories act effectively
- We are seeing systems fragmented and moving at greater speed
- Humans need to create something like coherence
- Trying to develop context

We can't understand how people make sense of things without understanding the system

Have a steady stream of info.
- When something happens - collapses our focus
- Then we dig back through evidence

We are looking for a way to predict and understand what MIGHT be happening.
- Currently inhabited by learning analytics
- Can't manage info flow through our existing social systems

We have hit peak social.  We are at point of diminishing returns.

It's not only about crisis situations.  Often more mundane.
- We are involved in sense-making activities on a daily basis.

We are at a point in human history - human cognitive capacity is lagging behind what we need to do / solve

Why analytics and what do we want it to do?
- Exceptional performance requires a tight coupling of people, roles, systems, context and actions
  + not haphazard

- Our socio-technical systems are becoming more technical and less social

- Important things are being lost in the "mass of the inconsequential:
  + We still need control
  + But the points are different than where the education system has assigned them in the past

How can centralized aims be achieved through distributed means?

INternet / mobile web fragments but to act we need coherence

- In response - we have become more rule based
- Social systems don't nscale with information's abundance / complexity

 Technological Society

Analytics - reduce serendipity
Technology creates more problems that technology can solve

SOLAR - Society for Learning Analytics Research
What kinds of questions do we need to be thinking about when applying learning and analytics to a social process?

Remember:  these are systems.  With human and technological agents.

Huge change in data-crumbs
- Activities tracked more
- We are happily sharing everything! Voluntarily!
- Constantly sharing of self

It's not even what YOU say is consequential
- Wouldn't take someone long to learn about you
- Through who you are connected to...
Data trails not just what we produce.  Our networks create data trails too.

Learning analytics - we are trying to optimize the learning for the individual student

We are probably going to enter a "learning stream"
- As the system figures out what we need to learn - the learning stream will send it to us

Learning analytics - sits across different spaces.
- Institutional research / IT / Faculty / Marketing
- Each has a model
- Cross-discipline concept (innovation in the gaps....hmmm.....)
- No experts (yet) - but some span boundaries better than others

Example - Quantified Self
  + Self-empowered, self-motivated learner
  + A guide to self-tracking

Any useful analytics activity - should cause you to ask more questions.
- Is there something actionable?

What if the topics of interaction are layered onto social networks?
- Nature of interaction
- Total Information Awareness

Distributed, multi-level analytics
- He does nothing with analytics in LMS - not a fan
- Drawing from a variety of systems
- Attention metadata - what we focus / spend time on valuable
  + Can start to tie this to time / date / who talking to
  + More info by being able to dig down in multiple ways

How do different kinds of media create different social network structures?

Tough to get a sense of what is in there and what we can automate.
- Automated vs. human analysis of data
- 70-80% accuracy between 2 different approaches
   + this is good for automated analysis

Learning Analytics - getting to hype stage.  Interest ramping up rapidly
- See Educause, Sloan events, SkillSoft Perspectives etc
- This will be a more substantial impact than eLearning

Participatory learning and reputation - see Yelp etc
- In learning space
- Increase reputation based on activity and engagement

Early warning - students at risk of failing
- Through analysis of social networks
- Signals project - analytics for students, better outcome with feedback

Analytics in workplace - not just class-based learning
- Analyzing also informal learning

Deloitte - can marketing data predict life spans

Colliding ideas

MOOC - half-ideas colliding to form new (innovated) knowledge wholes.
Dissolve boundaries of communication cycles.
- We fill each other's knowledge gaps

Creating: artifacts, stuff, remixing, new assemblies, novel connectedness
- The way you develop a coherent view of a situation - need to engage with it

Using gRRShopper -

We can't control network stuff in the same way we controlled things in the network era.
- Apply intelligence - stitch together the different pieces

Stigmergic - self-organizing social systems
- People can synchronize and align with people around them
- limited capacity long term

Synchronization

 How social systems right themselves when something out of synch

What might we want to predict?
- Composition of teams
- Skillsets needed for outcomes
- Resilience and adaptive teams in unsettled contexts
- Trends (for re/up skilling)

How do you "reskill" millions of people
- try to optimize

 Storage - we may be looking at fragmented warehouses
- Harmonize without Normalize -  LinkedData

- The END USER is the one who does something with it;.

All of this is prefaced on the ability of the individual to make sense of the data
- Data only goes so far....
- How do you DO SOMETHING with this?

Mediators do not need to apply
- Give us the right tools and we can figure it out

Systems - LMS
- initially, tried to lock down data.
- now - seeing LMS companies opening data.
- doing this if you can lock analytics into their system - will lock the org.
- Locking through the Data analytics process.  (Beware)

In networks - it's harder to control socio-technical systems
- Rule: You can't do this.  Reality: they will do it anyway

Groups doing stuff well - health care is starting to do this well
- If you can predict how people will become sick - really valuable

Other groups
- education bubble a pinpoint
- IBM and Google doing analytics well
  + Google may be taking analytics too far
 - Financial - esp eTrade. Does a nice job presenting to us

University - Purdue is a good benchmark
- Altius Ed
- The LMS vendors are very invested (since this is where they will make their money)


Other supplemental material (and because RSA Animation is cool)

-
www.youtube.com/watch?v=NugRZGDbPFU


No comments: