Algorithm Arrogance at Facebook

Pope Paul V - wikipedia, portrait by Caravaggio | https://en.wikipedia.org/wiki/Pope_Paul_V#/media/File:Paul_V_Caravaggio.jpgPosted to a Marketplace report on the most recent content stream tweak by Facebook:

It’s algorithm arrogance. There are many data science specialists working at Facebook, but there is reason to believe the new stream tweaks will not improve appreciably. One reason: users have no way to designate content you *do not* want to see (perhaps ever). Another: Facebook search is so unfriendly that search is rarely used to discover what you *do* want to read. (It’s part of the ever-popular toilet paper roll user interface). In other words, there’s plenty of data but not enough of the right sort to improve personalized relevance. Sure, not everyone would use a recommendation / search facility, but for those who do, the results would improve. The data “science” folks have become so algorithm-arrogant that you’d be hard pressed to even find a resource to personalize and improve your feed — with more data.

Best of Global Breed: Healthcare Systems Surveyed through a KPMG Lens

Photo: In Search of the Perfect Health System by Mark Britnell (book cover http://bit.ly/1GX8ktc)

In Search of the Perfect Health System by Mark Britnell (book cover)

[schema type=”review” url=”knowlengr.com” name=”In Search of the Perfect Health Care System” description=”Commentary on a book by Britnell surveying health care systems” rev_name=”In Search of the Perfect Health Care System” author=”Knowlengr” pubdate=”2015-11-11″ ]

Unhappiness over the Affordable Health Care Act (ACA / Obamacare) comes as much from the left as from the right. To learn what was right about the ACA, and how US healthcare being done better elsewhere. a book by KPMG’s Mark Britnell attempts to look at the global big picture — using an IT analyst-like “best of breed” survey.

In Search of the Perfect Health Care System (Palgrave Macmillan, 2015) gives that a try.

The publisher’s synopsis:

Have you ever imagined what a truly great health system could look like? Over the past six years, author Mark Britnell has worked in 60 countries – covering eight-tenths of the world’s GDP – with hundreds of government, public and private healthcare organisations.  With chapters on 25 different countries, including Brazil, China and the USA, his practical, succinct guide to the world’s major health systems explores what lessons can be drawn from each to improve health worldwide.This insightful and informative exploration of health systems around the world will give you a truly global health perspective.

Fierce Healthcare’s editor summarized the twelve components of Britnell’s book in a recent editorial as: universal, emphasizing primary care excellence, community level services, mental health, patient self-responsibility, health promotion, R&D, research diffusion, IT support, choice, patient empowerment, effective funding levels, elder care.  Paraphrasing is mine.

Glad to see patient empowerment on the list, but the list length reflects an unavoidable complexity.

Britnell was interviewed by Leonard Lopate in November 2015.

Big Data and IoT in Sports: Forecast Come True

Screenshot CIO Story on IoT in NFL by Thor Olavsrud http://bit.ly/1Zio3ZxIn a blog post written in January 2014 at Syncsort.com (“Big Game, Big Data: How Football is Being Transformed by Big Data”) I forecast that Big Data and the Internet of Things would eventually impact major sports in the U.S.  In a feature story written for CIO magazine by Thor Olavsrud (@ThorOlavsrud) in September 2015, parts of this forecast may becoming reality for the National Football League.

Question: How will it affect your bets in fantasy sports?  IBM Watson for unstructured expert advice? QlikView and Tableau for analytics?

Reviewing Peer Review

Screenshot of Retraction Watch post: http://bit.ly/1M0tJO8Peer review is thought to be the gold standard for advancing “proven” science, but those who regularly publish and act as peer reviewers know that peer review has its problems. A recent study published in BMJ Open looked at this issue, and was the subject of a post on Retraction Watch. I posted a somewhat lengthy comment which addresses some broader issues that have surfaced in my work with the Elsevier-sponsored Innovation Explorers group.

Chasing Big Data Variety: Predictive Analytics, Meet Your Market Foe

 

Linkedin Stock Price Graph - Yahoo Finance via Google Search 20150430 (screenshot)

The graphic shows the market behavior of LinkedIn’s stock price late afternoon of 2015-04-30. Did your analytics engine (What’s an analytics engine? See International Institute for Analytics) predict this? If not, what (big?) data were you missing?

If not, chances are, yours was a Big Data Variety problem. Correlating with, for example, only Facebook, Pinterest and other social media platforms may have been a tipoff, but not enough to forecast a 25% single day plunge.

And before you reach for the “Sell” button, you might want to revisit this two-year-old story on Forbes, when the stock price also fell. Did your analytics take that into account? The loss was less dramatic, but the cause was similar.

You may need data from other sources, and more than just sniffing URLs from corporate PR departments a la Selerity. Perhaps your forecasting engine treated that as just a day’s or a quarter’s data point, without consideration of the underlying cause. A mix of complex event processing combined with other types of machine intelligence might have had better results.

Why Computers (and Doctors) Need Narratology

Image of the StoryTellers Cafe

Image Credit: Loren Javier | Flickr

The analysis by Peter Kramer @PeterDKramer in the New York Times story “Why Doctors Need Stories” points, in part, to the challenge faced by clinical decision support systems (CDSS) — and the use of artificial intelligence in health care more generally. While CDSS adoption lags far behind its apparent value, it is true that CDSS is weak when it comes to sense-making from narrative. The latter is still a subject of much research in cognitive psychology, with much work remaining to be done. The widespread familiarity with machine learning and keyword search perhaps hides the importance of vignette-driven inference. And the point should probably apply beyond health care to other software-assisted analytics. Therein is to be found the real human role as knowledge worker.

IBM Watson? Work on your narratology.

Computer-Assisted Instruction: Long Overdue in Overburdened STEM Classrooms

Scientific American Story from 1978

Scientific American Story from 1978

This weekend the NYT is running a compassionately told story about challenges facing teachers in traditional classroom settings. I truly felt for the novice teachers.

On the other hand, having been classrooms with unruly students — especially with STEM-heavy curricula — I felt that even the tricks, tactics and cajoling described here would often fail  with some of the children. A continual return to behavior management, while necessary, is nothing short of a content pause for attentive students. Worse, in the absence of computer-assisted learning, the ability of a single teacher to track learning needs for component skills of, say 25 students — inevitably different from learner to learner — is spotty at best.

Teachers are implicitly asked to supply this missing computation in the form of homework grading, longer hours, self-produced content. What sort of person would want to accept such an assignment – with some nurtured while others just muddle through with at best intermittent attention to both subject matter & discipline? While the NYT reporter makes a good case for the importance of behavior management in the classroom, the narrative begs the question as to what works in behavior management and whether educational psychology pedagogy in teacher training is up the task.

 By analogy, what is it like to work in a factory where a constant number of products shipped are known to be flawed?

 One can hardly blame teachers for a system that begins more as child care than instruction. That it remains so for classrooms with older students here speaks to the Sisyphusian nature of the endeavor.

Image Credit: Scientific American story from 1978 provided by Steve Eskow in 2009.

Book Review: Business Storytelling for Dummies

business-storytellingHomer’s Odyssey was preserved through an oral tradition of storytelling. NPR listeners are familiar with StoryCorps, a show that features some of the 45,000 interviews recorded by the organization of the same name, and The Moth, another show that features unscripted storytelling.

At the opposite end of this humane tradition is “Death by PowerPoint.” This critique of the slide-based software, so named by Angela Garber, argues that PPT decks are often boring, oversimplified, needlessly stylized, overly complex – or all of the above.

Somewhere in the middle, perhaps, is ordinary business communication, instructional materials and well-tolerated military briefings. After all, David Byrne has shown in his PowerPoint-based art that the software medium itself may not be at fault.

So it is with PowerPoint in mind that I supposed that Dietz and Silverman’s Business Storytelling Business might be a useful mate (or antidote) to the book used for reference when the Microsoft PPT help screens fall short. Business Storytelling is written in a breezy style that belies the importance of this topic. Nonetheless, Dietz and Silverman have crafted a book that satisfied my goal.

I only own two “Dummies” texts, and the other one was printed in 1999, but the brand’s format hasn’t changed much despite the transition to Wiley. The text is strongly edited, by which I mean it follows a somewhat principled structure and doesn’t stray from its pragmatic intention for long.

A Must-Have for PowerPoint Mavens and TED Talk Wannabes

A psychological / linguistic term for storytelling is Narratology: “the branch of knowledge or literary criticism that deals with the structure and function of narrative and its themes, conventions, and symbols.” Business Storytelling doesn’t cover narratology in depth  (and fails to mention Roger Schank). While there are relevant and useful peer-reviewed references, they’re sprinkled throughout the text and interspersed with blog posts, anecdotes, books (e.g., Nancy Duarte’s Resonate: Present Visual Stories that Transform Audiences) and case studies. This is not a failing given the book’s stated audience and objectives, but the suggestions made would have been strengthened by an academic-style bibliography.

Offsetting this lack are numerous story examples – both good and bad – which illustrate the authors’ recommendations. Here are a few that may demonstrate the flavor of the book:

  • Identify a setting, characters, events
  • Build in empathy for the main character
  • Ways to include data in a story
  • Problems with insider and technical terminology
  • Create multiple layers for a story to reverberate
  • Ways to make a story memorable
  • How to demonstrate that the storyteller is simultaneously an effective listener

The book uses cartoon-style iconography for tips, pitfalls, realistic examples, external references, points to remember. Some may find this style off-putting, but if you can get past the idiomatic style, it does improve skim speed, which is what’s needed to rework and groom a story – and a PPT deck. Refer to additional worthwhile content at the book’s web site.

The business of creating business stories is non-trivial. Some of the unpersuasive examples provided in the book make this clear. Good storytelling calls for multiple talents: creativity, engaging diction, a balance of the simple and the complex, good aesthetic sense, practice, a sort of storytelling erudition – and, above all, metacognition of the domain that extends beyond the vanishing point of the story. Good storytellers may do all this instinctively, but, as a good fiction writer will readily attest, a cookbook will not result in a PowerPoint rose as lovely as Faulkner’s “Rose for Emily.”

You can see a TEDx talk by Dietz in which she introduces some of these concepts.

Feedspot: RSS Reader for Google Reader Diaspora

Feedspot: A New RSS Reader

Feedspot: A New RSS Reader

There was plenty of hand-wringing when Google announced that it was ceasing support for Google Reader. As is somewhat typical with Google’s project kills, it was a relatively precipitous decision that had analysts scratching their heads and users scurrying for alternatives.

Feedspot is one of several browser-based Really Simple Syndication (RSS) readers that offers features similar to Google Reader, and — this was critical for many users — includes Outline Processor Markup Language (OPML) import. OPML was the format used to export RSS feeds from Google Reader, as well as for other applications, including the venerable Microsoft Outlook.

As one of the refugees from Google Reader, I had many feeds spanning several engagements since 2005, I had a modest set of hierarchically structured feeds. Some of the folders were quite deep — e.g., “MusicTech” had 77 feeds, while others had just a few. I was a fairly energetic user of folders for structuring as well, and it is easy to imagine that others had more extensive lists. In total, there were around 750 RSS feeds. These were successfully imported into Feedspot in around 20 minutes in the middle of the day EST (US).

Feedspot has the feel of a relatively mature application. The small touches are part of this impression, such as the “tooltips.” An example is shown in the screenshot: “Unread article: A blue triangle in the top left corner indicates unread article. . .”  Other examples include the list-view toggle and the rightmost “Feedback” button leading to a UserVoice dialog. Feedspot also allows building of RSS collections — essentially, feeds of feeds, at the level of folders.

feedspot-list-switch

List View Toggle

feedspot-feedback-1

UserVoice Feedback

A so-called “modern” application must integrate numerous sharing opportunities, and Feedspot follows this trend. Features for sharing content are provided in several spots [sic]. Following of user collections is possible (“follow all of my stuff”) and the obvious — sharing of individual articles.

There are still some rough edges here and there. It was unclear how to place a new manually entered feed, place it into a folder, or how to create more than a 2 level hierarchy if that is supported.

RSS content is sufficiently important that I have placed my post-Reader bets on more than one pony.  Feedspot is one of them. It deserves serious consideration as a Google Reader replacement.

Founder on the Spot

I corresponded with FeedSpot founder Agarwal about the Company. He disclosed that FeedSpot was built using agile methods, and that it runs on a LAMP stack. His business goal? He says he always wanted to build “a consumer internet product.”  Why RSS? Because, he wrote, “for some users an RSS reader is a must-have product.”  But Agarwal believes it is possible to “take an RSS reader to mainstream consumers.”

Future of RSS?

The cancellation of Google Reader had some pundits predicting the end of RSS. Some consider the Facebook-dominated landscape as superceding the lowly RSS. The RSS button, they say, is losing ground to “Follow” and “Like” buttons. Indeed, there is a rich future ahead for the underlying data from those interactions. Still, it safe to say that pronouncements of the death of RSS are premature.  RSS is widely used in content management systems (e.g., Blogger, WordPress, Sharepoint and Joomla) to import links to relevant articles. This is true for externally produced content, of course, but perhaps it has even greater value as a rapid information aggregator for smaller scale intranets whose limited staffing prevents more sophisticated schemes.

RSS, partly because of its UserLand roots,  offers a simple but flexible framework for information management. It doesn’t deliver a directly usable ontology from an automation standpoint, but its wide adoption gives users great leverage to employ ad hoc schemes.

feedspot-uservoice-feedback-1-350px

UserVoice Dialog Box

The addition of tagging features in Feedspot strengthens the quasi-ontology capability in RSS. If judiciously used with a controlled vocabulary, search can be more fruitful. As longtime users of Gmail can attest, having to choose between folders and tagging was an uncomfortable either/or decision. Eventually Gmail would offer both for classifying email. Having this capability for managing RSS is helpful.

There is more fertile ground in the reference community, such as RSS feeds that are supported in ResearchGate, CiteULike and academic publishers. If feed schemas become more sophisticated, feed users could deploy them more rapidly. Content publishing based on feeds would more often reach appropriate reader communities. Content delivery at present is notoriously hit-and-miss, especially since bloggers can readily veer off the ostensible topic of the blog.

These are growing pains that have made the semantic web grow so unsteadily. But Feedspot and like tools are perfectly good hammers for the right kind of nail.

feedspot-showing-share-and-tag-circle

Tagging Feature

[schema type=”review” url=”http://knowlengr.com/blog-type/feedspot-rss-reader-for-google-reader-diaspora/” name=”Feedspot: RSS Reader for Google Reader Diaspora” description=”Blog site fror Krypton Brothers founder Mark Underwood. He blogs under the name “knowlengr” (knowledge engineer).” rev_name=”Feedspot” rev_body=”A review of the RSS reader Feedspot” author=”Mark Underwood (“knowlengr”)” pubdate=”2013-07-27″ ]

 

Analyzing the Beast that is Cybersecurity

Hyatt Regency walkway collapse (credit: Wikipedia Commons)

Hyatt Regency walkway collapse (credit: Wikipedia Commons)

What sort of beast is “cybersecurity” anyway?

Failure Analysis

Is it simply a variation of software failure?  According to this analysis, a security lapse is a software engineering failure, not technically different from an unintended “404” error or an “uncaught” exception.

Protection Analysis

Is it simply a failure to implement corrective measures? This analysis likens cybersecurity to physical security.  Facilities such as military bases or electric power plants are vulnerable targets. Rather than try to remove all the points of vulnerability, a virtual “layer” of physical security is drawn around the facility. The Department of Defense Physical Security Program provides a useful glimpse into this approach. Consider DoD 5200.08-R. A version last updated in 2009 is hosted by DTIC. To some extent, there is a reasonable analogy to protecting software.

Architecture Analysis

Is it a design failure? In architecture, it is not uncommon for architects to receive the blame for collapsed, or otherwise unsuccessful buildings. For instance, consider the failure of a walkway in the Kansas City Hyatt Regency hotel in 1981.  In this failure, 114 persons were killed, and initial blame settled on architects. A more nuanced view recognizes multiple sources of responsibility, including project sponsors, customers, auditors, and sometimes public officials and even politicians. This was the analysis made by one K. Bristol in a 1991 analysis of the Pruitt-Igoe towers project in St. Louis.

Regardless of which of these approaches is chosen, the relative contributions of alternative models for failure should be taken into account.  There is a tendency to focus excessively on the specific lapse (e.g., buffer underflow). Issues such as engineer training, IDEs, development frameworks, test environments and constraints imposed by sponsors and other stakeholders also deserve investigation, if not blame.