That is the goal of Azure Machine Learning. “This is, as far as I know, the first comprehensive machine learning service in the cloud,” Mr. Sirosh said. “I’m leveraging every asset in Microsoft for this.” He is also using ways of accessing an open source version of R, a standard statistical language, while in Azure.
Many of us expert in industrial recommender systems matching “S’pply n D’mand” missed the change from shortages of supply to shortages of demand. VCs didn’t miss this backing anything mining demand inferences, particularly Hadoop based leaving new sources of global supply from global suppliers by companies like Apple. In doing so the open source Web 2.0 hadoop business architecture was created to mine in closed "big data" sources for your demand: Results soldto advertisers.
Private Equity funds didn’t miss this shift from supply to demand, taking nearly every public ERP company operating at scale private (virtually all bar SAP and Oracle). At the core of ERP is MRP (material requirements planning) which business use to match supply/demand. With shortage of supply behaves as recommender system for exceptions. With abundant supply, quality demand becomes the imperative.
The VC’s backing quality demand brought us CRM, Search, Social and Ecommerce. Greg Gottlesman, VC at Madrona believes that Amazon out of all is in the process of becoming the world’s most valuable company.
Amazon’s businesses create the best quality demand, a credit card cleared order, not social engagement inferences mined from Facebook via Hadoop et al.
What can upset this Golden Pony, shortage of supply?
Shortage of water, oil and other precious raw materials can and will disrupt supply and demand. For example McDonalds and Nike are already investing in smaller suppliers and alternative sources to add diversity and reduce risks in regions of the world with sustainability issues like Indonesia and drought regions.
Nike even went as far as hiring Ward Cunningham to help with prototypes for working in the emerging era of sustainability. Alas, currently over investment in demand systems feels inversely proportional to under investment in supply systems technologies.
For example the core algorithm for MRP, around since 1970s has been revised for the demand driven era by Carol Patk & Chad Smith in Orlicky’s third ‘DDMRP’ edition — but not web 2.0 social technology updated. Internally we can surmise Amazon has built Hadoop MRP.
The DDMRP authors are superb APICs, Goldratt, TOC, Lean, operations experts. Not computer scientists versed in online recommender systems, machine learning, or social computing technologies (not Carlos Guestrin,Ward Cunningham or Ed Chi types).
What can upset this Golden Pony, shortage of demand?
Another possibility, the search for quality demand turns away from big data to personal data. Ward’s - Smallest Federated Wiki, Brad’s - Camlistore, and Doc Searls - project VRM all point in this direction.
The Respect Network is helping bring personal quality demand via emerging personal and enterprise datastores. These private by default, shared in a granular fashion, personal clouds were part of Doc’s keynote for the respectnetwork.com brings trust into perspective.
Other technologist’s such as Adrian Hall have startups such as deconstructed.io focused on user data flows across devices without mining for targeting.
Net-net here, I need to find an enterprise startup crazy enough to explore updating MRP here, matching demand/supply via VRM/Respect.
Jordan Ritter of Napster fame has set up Ivy Softworks for Data Psychics to incubate joining up the dots in this space. Till then, inspired by a16z’ Ben Horowitz, listening to the Egosouls-Nirvana: Produced by S’pply ‘N’ D’mand: Think I’m going crazy!
“Personalization appeals to a Western, egocentric belief in individualism. Yet it is based on the generalizing statistical distributions and normalized curves methods used to classify and categorize large populations. Personalization purports to be uniquely meaningful, yet it alienates us in its mass application.”—Data Doppelgängers and the Uncanny Valley of Personalization (via johnborthwick)
Editor’s note: This is the third in a periodic series profiling Microsoft workers amid the company’s sweeping attempt to foster a more collaborative culture, one that’s more innovative and agile. This story profiles Corporate Vice President Peter Lee, head of Microsoft Research.
For the longest time MSR produced cool new technologies that went nowhere. Product Managers bonused on Windows, Office and Server division sales blocked new features unless they could pull thru at least $200 million. This is how Microsoft ended up with rubbish adoption in mobile, ceded NoSQL to startups, missed out on big data with Dryad. The list goes on and on.
Your customers’ feedback is the most important asset you have when building your SaaS offering. Get it often, and make d…
My experience is weaning a growing software company off custom development to more sustainable and scalable product development.
The company had become trapped in sales driven feature requests creating huge technical debt due to using old tools and swinging to add features to win deals without a cohesive product plan.
It took some time and buy in from the exec team to drop grabbing every dollar or pound and focus on selling solutions we had not features we didn’t. At one point they nearly crashed the company promising features without a plan to build them. Still the founders would never have gotten venture funding without going the custom feature route. It’s just they kept on going for about 4 years before they should have pivoted to a product driven company, not a custom feature driven. For example deep in code IBM had special purchasing features that no one else could use.
Much later on Parker Harris, Co-founder, Salesforce.com, and Microsoft Project Green solved the custom SaaS architecturally, with huge success. I solved with an SDK / API and prepackaged sample code (both ate required, one is not enough on its own).
I am not foolish enough to say that the way I write software is the only way, but I am foolish enough to believe how I approach software design might be helpful. This isn’t about project management, or algorithmic strategies, but how I fumble around until something works, guided by the scars of…
How did you know how I design software solutions? ;-) Great post, thanks.
Great talk between Steven Sinofsky and Box CEO Aaron Levie about how the cloud is shifting the approach that vertical industries take to implementing systems and infrastructure. Some of the most profound changes are indeed happening in the deep recesses of highly verticalized industries such as medical device and building materials and diversified industries. The solutions provided by the software industry to date have either been too broad to work well or were too wedded to the industry to bring anything innovative and fresh.
The topic of API’s was brought up which is one part of the equation in building better vertical solutions. However, we would say that improvement first and foremost has to happen at each and every point of interaction with customers. CRM systems still struggle with this, mostly due to the user experience. By leveraging mobile and giving business people control of the functional layer of applications, companies can realize marked improvements in user and customer engagement, thus feeding into systems and building more comprehensive view of the customers and the overall business. But the key to all of this is giving business people that control, the ones that know and understand the industry needs best. That is exactly what Enhatch delivers for our customers.
“Technology leadership is not defined by patents, which history has repeatedly shown to be small protection indeed against a determined competitor, but rather by the ability of a company to attract and motivate the world’s most talented engineers. We believe that applying the open source philosophy to our patents will strengthen rather than diminish Tesla’s position in this regard.”—
… containing your medical records … oh alright, I’ll tell you.
The Riak documentation on security says (and I only slightly paraphrase) ” “write it yourself”. Which is the same as not having any really. But as far as I know they didn’t implement any anyway, so I suppose the product fitted their requirements perfectly.