Friday, April 3. 2015
Two rather interesting snippets on the veracity of Big Data analytics. Firstly, from HBR, this finding (summarised below):
Researchers recruited 61 analysts (mostly academics) and asked them to assess whether soccer referees were more likely to give red cards to players with darker skin tones. The analysts split up into 29 teams, and were given a dataset that included numerous variables about both players and referees.
With 29 teams and 29 slightly different results, it is clear that any analytical depends on somewhat subjective decisions about the best approach to use and which variables should be included. After another round of debate, the analysts "converged toward agreement that there is a small, statistically significant relationship between player skin tone and receiving red cards, the cause of which is unknown".
The second is also from HBR, again by Walter Frick (who wrote the above article), with research showing that even if the algorithm was right, most people tend to prefer their gut feeling, even more so if they have seen an algorithm fail, even a little bit. And they’re harder on algorithms in this way than they are on other people:
Initially this looks like irrational behaviour, but of course if you juxtapose it with the above research that shows that one should probably not trust any one algorithm, it makes sense.
Apparently most people, when asked why they trusted human judgement, felt that humans were better at learning over several iterations (this is not demonstrably true, by the way).
I recall being told 30 years ago when starting out with complex simulation modelling that "people would rather stick with a problem they can't solve than a solution they can't understand".
Interestingly, in a forthcoming paper, the same researchers found that people are significantly more willing to trust and use algorithms if they’re allowed to tweak the output a little bit. If, say, the algorithm predicted a student would perform in the top 10% of their MBA class, participants would have the chance to revise that prediction up or down by a few points. This made them more likely to bet on the algorithm
Which brings us back to 29 teams with 29 different models.....
Thursday, April 2. 2015
Some time ago I read a fascinating essay on the Economics of Social Status by Kevin Simler. It essentially looks at status as an economic good, and I found it an interesting way of thinking of online social capital (aka "whuffie"). Anyway, I was reading it again as background to some thinking I'm doing in new types of Organisation Structures" (see the Anna Karenina Principle, here) and I re-read this bit:
"A community is a group of people who agree on how to measure status among their members".
Whenever "Transaction Costs" are mentioned I think of Ronald Coase and the Theory of the Firm, which says that people begin to organise their production in firms when the transaction cost of coordinating work through market exchange, given imperfect information, is greater than within the firm.
On re-reading this, I think there is something profound here for online social networks, in that the transaction costs (a proxy for efficiency and effectiveness) of any social network organisation structure are fundamentally about the ability to trade social capital (a proxy for trust).
There is also something interesting here for all the many wannabe systems vying to replace the good old hierachical way of organising work - it's not the easy movement of information that is key, it is the easy movement of social capital - aka influence - that drives the benefits of more connected organisations. There is an implication here, that simplifying social capital markers (and that includes hierarchy) is key, which implies that heterachies, while fine for non-human systems, may be less effective as organisations for humans.
Food for thought.....
Thursday, March 26. 2015
An interesting day the SCTE Spring Lecture yesterday – thanks to those speaking and organising the event. The headline topic was DOCSIS 3.1, which is the next evolution of cable modem technology. I have spent the last few years working with DSL-based ISPs, so it was good to find out what’s been going on with cable.
DOCSIS 3.1 updates cable modem technology by increasing the potential spectrum available and improving the efficiency of is of spectrum use. It also opens the door to better integration between the TV data sides of cable technology, as well as standardising platform management interfaces. However, it seems the most important thing for cable companies is to get more bits through their existing infrastructure. This is good for the companies and the consumer.
Like any infrastructure business, telcos and cable cos want to squeeze their assets as long and as much as possible. If we had invented broadband access as a green field technology, we would have laid a fibre to every home. Of course, digging trenches is expensive and engineering should be the art of the possible. In the telco world, the DSL family of standards were developed to squeeze data over the twisted pairs originally designed for voice. (A quick historical note - ADSL was originally developed to carry video on demand over ATM, back in those seemingly distant days when the Internet was a curiosity for geeks and academics.)
A few years later, in the late 90’s, DOCSIS was developed to overlay data on cable TV networks. Because cable TV uses co-axial cable rather than twisted pairs, this was always an easier task. Cable does have a disadvantage because the cable is shared by many customers and therefore the bandwidth has to be shared. This is known as the “contention ratio”, but DSL suffers from shared resources as data goes deeper into the network and this is really the point of the Internet i.e. a shared medium that can be used by everyone intermittently.
Cable companies have dealt with contention and increasing demand for bandwidth by -
• allocating more bandwidth to broadband data
• improving efficiency (bits/hertz) by upgrading the technology to allow better performance over the same network.
• improving the network quality to reduce noise and so improve bits/hertz (also known as the constellation size.)
• segmenting the network to reduce the number of customers “sharing” each data feed.
From the mundane exercise of sending technicians to tighten connectors (improving network quality) to the high-tech of advanced line coding (new technology), these approaches all have costs and benefits and so it becomes one side of a business case. The other side is the competitive environment. In the UK, Virgin Media are offering 152Mbit/s as their top tier and one has to assume that this pitched so as to outstrip anything BT can do using DSL and twisted pair! In Europe, teclos are using Fibre to the Home (FTTH) to reach Gigabit speeds and the cable operators are responding. DOCSIS 3.1 will make this much easier for them.
As an aside, a question was raised about the theoretical limit to the amount of bandwidth a human can consume. We didn’t get a good answer, which was fair enough as it is really a question that involves cognitive psychology. However, it did bring to mind a review I read many years ago for a V.32 (9.6K) modem that said something to the effect that it was all very clever, but no one can type that fast!
Wednesday, March 18. 2015
Layers of News Media over time - Business will be no different(from Baekdalmedia.com)
We've written about this a few times on Broadstuff, but not for Social Business. I wrote about this application of Riepl's Law to Enterprsises over at the Agile Elephant Blog, but in short:
It’s never going to happen. Email will be here for a long time still, so get used to living with it. Social Business systems that can’t cope with email will die a long time before email will.
The reason for this is that, in the entire history of new media from the invention of speech onwards, newer and further developed types of media never replace the existing modes of media and their usage patterns. Instead, a convergence takes place in their field, leading to a different way and field of use for these older forms. The diagram above shows how mewdia generations have gone in News, it will be no different for Business communications. (Source Baekdalmedia.com)
This observation is called Riepl’s Law.
This was first noticed by Wolfgang Riepl. Riepl was the chief editor of Nuremberg’s biggest newspaper at the time, and was stated as above in his dissertation about ancient modes of news communications.
Tuesday, March 10. 2015
News this week that GigaOm and FriendFeed have closed down.
GigaOm was one of the (better, IMO) Tech blogs-cum-digital news sites, the problem was (i) so many others also opened up at the same time, and (a salutary lesson) GigaOm was more about sound analysis rather than froth or shilling. If there is one lesson from Digital media it is that heavyweight content usually sinks, we are still largely in the "digital weeds" phase of ecosystem development in digital media . TechCrunch* soild itself to AOL, this does raise questions of whether there will be a larger shakeout soon.
Friendfeed was one of the (many) experimental social network approaches trid out in the 'noughties - some modes like Twitter & Facebook "stuck", many didn't. Joost, Seesmic, Plurk....remember them? The Friendfeed founders were smart, and Google bought Friendfeed for its people, but the product was left (as so often) to wither slowly on a decaying vine.
In reality though, the two are merely part of the huge evolutionary process going on as part of the overall Digital Transformation into Homo Surfiens, no different to the huge experimemntation of say the shift of shipping from sail to steam, or development of aircraft, or any other radical new technology you can mention. A huge number of experiments are tried in the Darwinian soup of technology evolution, some climb up the slippery stick and make it as "best ways forward" and jump the chasm, the others are subsumed into the mud silt at the bottom of the evolutionary ecosystem as experiments that failed....till next time.
Ashes to ashes, dust to dust, startups to silt.... but the overall Digital Transformation continues apace. It's often salutary to think back on the world 5, 10, 15, 20 years ago and look at what has changed in ICT as Moore's law carries on carrying on. And just think, if it wasm't for Edison we'd be surfing our tablets by candlelight....
*Update - I note its founder, Mike Arrington, tweeted today that he was glad he never took outside funding for TechCrunch. Therein probably hangs a tale. To paraphrase Tolstoy, all succesful companies are the same, every failed company fails in its own way.
Monday, March 9. 2015
I published a summary of the Lecko report on Social Business is France 2015 over on the Agile Elephant blog, in two parts:
- Research into the evolution of deployment and what companies are doing
There is quite a lot that is useful for UK and other countries rather than being just France specific. The most powerful part of the work however, in my view, is the analysis of 35 software packages against c 550 datapoints (see analysis diagram of one package, Jive, above), its a very good way of comparing for selection and "best fit" analysis, also allows them to make a much more transparent "Gartner Quadrant" estimate (see below)
Tuesday, February 10. 2015
Bjoern Negelmann and the Gartner Hype Curve of Social Business
I wrote up 2 notes in our Agile Elephant Blog (over here and here) on the state of Social Business today, but here is a summary:
In essence I buy Bjoern Negelmann's analysis at the Enterprise 2.0 Summit in Paris last week that it is in the Slough of Despond phase, and the only way is up (see picture above).
This is confirmed in my view by the latest McKInsey report on the subfect, which essentially notes that it is here to stay, and more and more companies expect to expand its usage.
I also think the fact that two "New New" memes - "Digital Transformation" and "Enterprise Social Networks" are now riding on its coat, tails tells you that it's hitting early mass markets. But last word to the McKinsey analysis of "what comes next":
Begin with a targeted approach, then broaden impact. While the overall adoption of social tools remains widespread, the results indicate that most companies use them intensely in only a few functional processes. Yet the successful use of social in sales-and-marketing processes suggests how much more potential value is at stake in other parts of the business. To get the most value out of social technologies, companies should focus on specific cases where these tools could be implemented in a targeted way. A company already using social tools could broaden the technologies’ impact by adopting them in areas such as operations, where they are used less often now.
From the Slough of Despond, the only way is up.....
Saturday, January 31. 2015
The British Army has resurrected the Chindits, one of the special force units used in WW2 to operate behind enemy lines - in their case to operate behind Japanese lines in Burma. It was formed by Brigadier Orde Wingate following a successful usage of Boer commando style tactics in the British/Commonwealth reconquest of Italian East Africa in 1940.
The new Chindits are to fight behind the lines in the digital wars - Torygraph:
...the new unit's focus will be on "unconventional" non-lethal, non-military methods such as "shaping behaviours through the use of dynamic narratives", an Army spokesman said.
It's part of the (belated, in my opinion) realisation among Western armies that from the end of the Korean War onwards, asymmetric warfare is the new "conventional" warfare. The new Chindits are designed to combine "conventional" irregular warfare psy-ops and disinformation techniques with a major digital capability to use social media and other digital information technologies.
One thing to keep in mind - the Chindit experiment ultimately failed militarily, their raids were characterised by dysentry wracked men carrying huge loads marching uselessly for days without contacting any enemy, high casualty rates due to disease not enemy action, and US (air) support was essential to keep them in the field*. They were retained more for PR purposes on the home front than any military effectiveness.
One hopes that the same will not happen now.....
(*This is not a criticism of the brave men who served in the Chindits, it's that an approach finessed for dry African conditions ultimately did not work owing to the jungle environment)
Friday, January 16. 2015
Google to stop making Glass "In present form"
Google is to stop producing its wearable technology Google Glass in its present form, but is still committed to the idea of smart glasses, the company has said.
In a statement posted to its Google+ account, the company said the Glass team would move out of the “Google X” incubator labs and become a separate division. What the "New New" form will be is anybody's guess, but I'd suspect it will be less obvious to other people more than anything else, to stop both the "Glasshole" problem and to not alert other people so visibly that its filming and recoding them. Somehow I don't think they will relinquish the "hoovering up all that data" play.
In other news today, UK hypermarket chain Tesco has just launched an app for Google Glass...talk about timing. While the aim is largely laudable, it seems like yet another unfortunately timed play from this beleagured retailer.
Update - Had a Twitter discussion with Steve Bowbrick, his point being that the Google Glass busiiness model (very expensive device that records other people as UGC) has been rejected (by the public). I am less sure that Google will let go of it, as with Facebook I suspect it will be a 1 step back, 2 steps forward when you're not looking sort ofplay - I think it will be interesting to see what 2.0 does.
Thursday, January 15. 2015
The next edition of Charlie Hebdo went on sale on Wednesday, and the production run was up from 60,000 to c 5 million - and rather than c 25% of 60,000 copies being remaindered, this time 5 million was probably not enough. (The money will be going to support the families of those killed, though there is a bit of a sour note starting with people trying to sell copies on eBay)
If the aim of the shootings was to cow Hebdo and its satire it has been a complete failure so far, and also far any attempt to further inflame France (if anything the UK media seems more cowed than the French).
But (and this is why the subject surfaces on this digital tech blog) modern online technology probably has quite a lot to do with this beng different to what it may have been even a few years ago. In short, my thesis is that the huge amount of user generated content has both increased ordinary citizen participtaion globally, and lessened the ability of the various vested interest armed camps to twist this to various agendas. My thesis is that, (so far anyway):
Firstly, using violence or banning to suppress unwelcome media is hardly a new trick, and it usually works if there is no way to route around it - but if it can be routed around it often produces the opposite effect, i.e. drives a far larger distribution and hence support base for the victims. (This is why banning stuff always backfires if the material can find a route out, as it incites more, not less people to take an interest). Digital media, including video and image based media, was a major part of that alternative channel here.
In addition, the shootings have generated furious discussion about a multiplicity of conflicting issues to a new level of distribution and intensity. There are many interlocked issues - Freedom of speech vs Causing needless Offence; What observances can a religion rightfully demand of non-believers; Is this religious fundamentalism or a cynical use of religion for political ends, What is the role of Satire - to only speak truth unto power, or to poke at any sacred cows, etc etc etc.
It has engaged millions of people all across the world like nothing I've looked at online before, and (I can't prove this on my very limited analysis to date, however I think this is true based on what I've looked at so far) I think we're not merely seeing the all too usual polarisation and a spit-at-your-camp approach that contentious issues usually take, there is actually some hope that we are still in an Age of Reason.
At this "interesting" time in history this is something that can only do good in this writer's opinion....
More Broad Stuff
Poll of the Week
Will Augmented reality just be a flash in the pan?
Creative Commons Licence
Original content in this work is licensed under a Creative Commons License