IIoT analytics and cloud developments this week

Well, this week your analytics roundup is distinctly more cloud than analytics I fear.  Nevertheless fearless readers, I hope I've found something enthralling for you, so read on....

• Dropbox bails out of Amazon’s cloud.  It's very easy to write about the tremendous growth of the cloud.  If you read my stuff regularly, you'll know that I've done just that.  Less common are those stories where a company uses the cloud, then does a U-turn and brings it all back on-premise.  Dropbox is doing that.  Also interesting to note that Dropbox is eschewing open source, another technology that is widely praised and also widely adopted.  Because as Akhil Gupta, VP of Engineering notes "...there’s nothing in the open source community that’s proven to work reliably at our scale.”


• Gosh, it really is easy to find positive stuff about the cloud, isn't it...?  This article looks at the accelerating adoption of technology (kinda) by looking at the market capitalization of traditional, dotcom and cloud era technology companies.  The key point is that cloud based businesses are growing their market cap at the expense of traditional vendors.  Dotcom era companies are basically holding steady.


• Did I mention that I had more cloud in this post than analytics...?  It was 10 years ago last week that Amazon introduced it's first cloud service.  From a simple storage service offering a whopping 5GB, Amazon Web Services has kicked on a bit.  This 10th anniversary retrospective from The Register is about as irreverent as it gets.


• But wait, I found some analytics stuff to write about...OK, it's machine learning, but that's close enough.  You may have heard that a machine learning program called AlphaGo was taking on a highly ranked actual person, Lee Sedol - a professional Go player.  Humans essentially acknowledged that they no longer held the upper hand at chess after Deep Blue beat Gary Kasparov in 1997.  Never mind though, Go is much tougher than chess and so we would reign supreme there for...oh, about 19 years actually.  As reported, AlphaGo soundly beat Lee Sedol, 4-1.  So, on the face of it, machines left us floundering again.  But wait, that's not the full story.  As Dae Ryun Chang points out in Harvard Business Review, some of the limitations of machine learning were abundant.  And as Lee Sedol is quoted in The Economist, “The typical, traditional, classical beliefs of how to play—I’ve come to question them a bit.”


• Let's finish this week with a bit of machine learning for the people.  A new startup called DimensionalMechanics is aiming to provide machine learning services that can easily plug into other enterprise applications.  Details are a bit sketchy at the moment, but the intent seems to be to cut out one of the more difficult parts of the machine learning process.  Namely, finding people who are skilled enough to build appropriate machine learning algorithms.