November 28, 2015
There’s a long interview with Stephen Wolfram in “Interview with Stephen Wolfram on AI and the Future,” which I found when pruning my archives. Here’s one of the quotes I noted:
Recently, computers, and GPUs, and all that kind of thing became fast enough that, really—there are a bunch of engineering tricks that have been invented, and they’re very clever, and very nice, and very impressive, but fundamentally, the approach is 50 years old, of being able to just take one of these neural network–like systems, and just show it a whole bunch of examples and have it gradually learn distinctions between examples, and get to the point where it can, for example, recognize different kinds of objects and images.
Hmm. Half a century. Progress comes from faster chips and clever implementations of well known methods. Interesting.
Enterprise search is also old. Improvements have been slow and seem to be lagging behind other fields. Is it the vendors or is the nature of the problem? The self appointed experts, failed webmasters, and former middle school teachers now working as taxonomy experts are pitching governance, semantics, and assorted packets of artificial butter.
Stephen E Arnold, November 28, 2015
November 5, 2015
Here’s a delicious statement attributed to Mathias Doepfner, an Axil Springer big gun. For context, remember that Axil Springer bought the Financial Times for something north of $1.3 billion. Now the quote:
“A takeover of this size … is neither planned nor foreseeable.”
Maybe an impulse buy like Google’s purchase of Motorola? I assume that Volkswagen did not plan or foresee the consequences of its emission control software behavior. German management approaches certain decisions in an interesting way: Neither planned nor foreseeable.
Stephen E Arnold, November 5, 2015
October 24, 2015
I read “FTC Chair Edith Ramirez Outlines Concerns about Big Data.” The write up contained a stunning quote to note; to wit:
The agency is concerned about “algorithmic transparency,” she said, and how “algorithms can be manipulated.”
Wong asked about connected health devices. “We are very concerned about this,” Ramirez said, adding: There is this flow of health information that is happening outside of the regulatory space. We want to make sure that very sensitive information is being protected.
Interesting. Will vendors reveal what their numerical recipes are doing? A better question: If the vendors revealed their algorithms, would most people know what the algorithms were doing? Asking for transparency and understanding are, in my opinion, two different activities.
Stephen E Arnold, October 24, 2015
September 9, 2015
Tucked into the business section of the September 7, 2015 New York Times was a story with the not too SEO friendly title “Competitors Say Google Is Slowing App Installation.” The write up is not indexed in Google News. I just checked.
The short item talks about mobile ads for apps. The article recycles a Google blog post which talks about penalizing “please, install our app” ads. Yikes. Penalties.
But for me the important item in the article was this quote, attributed to a Xoogler named Mike Dudas. He said:
This seems like a really strong move in deterring people from installing apps.
Ah, freedom. I too enjoy the smell of napalm in the morning. Wait. I have a question? Who is the enemy? Android centric developers? Do I hear the thump thump of choppers?
Stephen E Arnold, September 9, 2015
August 16, 2015
I read “News Corp’s CEO Bizarre Obsession With Made Up Lies About Google.” I am fascinated with “real” journalists approach to information. Here’s the passage I noted:
That Google’s newly conceived parent company is to be called Alphabet has itself created a range of delicious permutations: A is for Avarice, B is for Bowdlerize, through to K for Kleptocracy, P for Piracy and Z for Zealotry.
The Alphabet Google is making news without the fancy words. I am not sure what kleptocracy means. More interesting to me was this item, which I assume is true: “Google Skirted Drone Test Rules by Using a Deal with NASA.” Google’s approach seems more efficient than other firms’ methods.
I do like that kleptocracy thing, however.
Stephen E Arnold, August 16, 2015
May 16, 2015
Short honk: I read “Intranet Search? Sssh! Don’t Speak of It.” It seems that enterprise search is struggling and sweeping generalizations about information governance and knowledge management are not helping the situation. But that’s just my opinion.
But set that “issue” aside. Here’s the quote I noted:
The only way this situation [search is a problem’] will change is with intranet managers stepping up to the challenge and telling stories internally. The problem with search analytics (even if you do everything that Lou Rosenfeld [search wizard] recommends) is that there is no direct evidence of the day-to-day impact of search.
Will accountants respond to search stories? Why is there no direct evident of the day to day impact of search? Perhaps search, along with some other hoo hah endeavors, is simply not relevant in today’s business environment? Won’t more hyperbole filled marketing solve the problem? Another conference?
The wet blanket on enterprise search remains “there is no direct evidence of the day to day impact of search.” After 30 or 40 years of implementations and hundreds of millions in search development, why not? Er, what about this thought:
Search is a low value utility which has been over hyped.
Stephen E Arnold, May 17, 2015
April 16, 2015
Here’s a keeper for my quotes to note folder. The source is the New York Times, April 16, 2015, page 8 in the business section (where else?). The article has the Google index friendly title: “Challenge to Google. Innovation May Undercut Case. As in Microsoft v. Europe, Innovation May Undercut Case.” Beefy. [If the headline disappears along with the story, speak to someone other than me. You may be able to purchase a dead tree version of the newspaper if you live in an area where distribution makes it available.]
Here’s the quote attributed to the astute Warren Buffer of search engine optimization expert, Daniel (Danny) Sullivan:
You don’t expect the New York Times to carry a rival sports section. But you do expect it to have a sports section. When people go to a search engine, they’re looking to search across everything.”
An interesting generalization. I am not sure most people “go to Google.” Most people use what ever system is baked into their “user experience.” But that’s less important than the suggestion that the “you” is what I do. Nope. I also noted the “everything.” I find that suggestion of comprehensiveness amusing. Not as chuckle worthy as IBM Watson’s smart software writing recipes, but it is right up there in the search Top 100 silly generalizations.
The quote also brushes against a larger question, a question ignored by the New York Times; to wit:
Why is ripping off the LA Times’ sports section egregious and taking other outfits’ digital facts sort of okay?
I think the answer resides in the little appreciated patents issued to Ramanathan Guha and Alon Halevy, both smart people and both Googlers when each did quite prescient work. If you are not familiar with the notion of building a comprehensive global knowledge base populated by nifty Georgia Tech T shirt wearing software agents, you are missing some useful color on Google’s assumptions, systems, and methods.
But, hey, who really cares about the global knowledge base thing, the notion of dataspaces, and Guha’s vision of applied semantic technologies in his quick slick architecture of smart software?
The Guha Halevy work is important in my opinion. More than generalizations will be needed for experts and legal eagles to figure out that the Google processes have been humming away for many, many years. Still few outside of Google understand what’s up?
Roll out the generalizations. Google will have a barrel of fun. Ignore penetrating questions. Google will rake in the dough.
Stephen E Arnold, April 16, 2015
February 4, 2015
Quote to note: I read “Google Brain’s Co-Inventor Tells Why He’s Building Chinese Neural Networks.”
Here’s the quote I noted:
Deep learning algorithms are very good at one thing today: learning input and mapping it to an output. X to Y. Learning concepts is going to be hard.
I support research projects. I don’t support marketers who ignore fact for hype. Also, note, that the co-inventor Andrew Ng jumped from Google to Baidu. The human brain can figure out compensation and opportunity I suggest.
Stephen E Arnold, February 3, 2015
September 28, 2014
Here’s the quote I noted:
“[Co-founder] Larry Page is an apolitical genius,” Assange admitted. “His ambition is a megalomaniacal ambition, but it’s the same ambition that many CEOs have, which is just to make Google as big as possible . . . [Page] is constructing this giant machine . . . but it has no color. It’s like white rice . . . and it will adapt to any flavor put on it.”
I don’t agree. Mr. Page is worried about money. Yep, I know that sounds crazy. Also, there seems to be some tension related to Google from some corners of the world, not Harrod’s Creek, of course.
Stephen E Arnold, September 28, 2014
September 26, 2014
I tucked this paragraph in my keepers’ file:
Yes, there’s plenty Google does that raises concerns. The growth in showing direct answers taken from other sites, the inability for people to understand if they were hit by an automatic penalty, the failure to update some of its automatic filtering systems in a timely manner are some of these. And Google deserves criticism over them. But News Corp has a long history of lashing out at Google in ways that exaggerate some concerns way beyond reality. For a media giant that could take a real leadership role, its latest more-of-the-same letter is a disappointment.
The source is “Dear News Corp & Google: An Open Letter On Their Open Letters To Each Other Over Competition.” I find this fascinating for three reasons:
First, I chuckled with the notion that search engine optimization helps explain the actions of two hyper capitalist operations. (SEO is one business focused squarely on manipulating precision and recall when one runs a search. Hey, this is the brave new world of search. The customer’s information need is secondary to sizzle and ad revenue.)
Second, the squabbling of actual / would be monopolists is a reprise of the Vanderbilt – Gould dust ups. Does anyone remember what happened? The fellows teamed up. Could this happen again?
Third, both companies take a dim view of nation states’ strictures unless it is to their advantage. The approach is, it seems, situational. However, when a government can disadvantage another operation, the law is truth and beauty.
Stephen E Arnold, September 26, 2014