Comment by hooande | Hacker News Reader

Comment by hooande | Hacker News Reader

Parent Full thread

What bothers me is the triviality of the examples in the article:

Next year’s version of the Excel spreadsheet program, part of the Office suite of software, will be able to comb very large amounts of data. For example, it could scan 12 million Twitter posts and create charts to show which Oscar nominee was getting the most buzz.

A new version of Outlook, the e-mail program, is being tested that employs Mr. Horvitz’s machine-learning specialty to review users’ e-mail habits. It could be able to suggest whether a user wants to read each message that comes in.

Google is building cars that drive themselves. Microsoft is still analyzing real time data and improving email.

I was hoping to hear about sweeping changes and products that would affect all of our daily lives. I still believe that the OS itself could use a lot of optimization and personalization. A truly adaptive information management system could be a real edge for any company in the OS space.

It could be that the author cherry picked examples that were easy for the average reader to understand. If not, another example of companies that are really big thinking really small.

> Google is building cars that drive themselves. Microsoft is still analyzing real time data and improving email.

What makes you think one is important but not the other? Google is building cars that drive themselves but they are also analyzing real time data and subsequently improving their email. It will be stupid for Google to ditch Gmail and direct those engineers to work on cars.

>It could be that the author cherry picked examples that were easy for the average reader to understand.

Well since NYTimes has a broader demographic than just IT professionals I think the examples do fit perfectly in.

> What makes you think one is important but not the other?

Because Google already did that and is doing other, more radical, things?

I was also disappointed by the article, but I think it was the article. Things like email prioritization (aka priority inbox, so google and microsoft do share something there) are hard problems, and shouldn't be trivialized any more than someone making huge advances in UI and ease of use should be, but I was hoping for some hint at long shots as well.

> It could be that the author cherry picked examples that were easy for the average reader to understand. If not, another example of companies that are really big thinking really small.

I think it might be partly that and partly the author not understanding the subject well enough. He apparently writes on the bits blog, and some of his other articles don't seem too bad, but this example didn't really do much for my confidence in him:

Microsoft is hardly alone among old-line tech companies in injecting Big Data into its products. Later this year, Hewlett-Packard will showcase printers that connect to the Internet and store documents, which can later be searched for new information.

If we're charitable, they only thing I can come up with is meaning storing a very large corpus and searching over it, but put like that, it isn't very cutting edge, and what do printers have to do with it?

and 12 million tweets is hardly "big data"

To be fair, microsoft did kinect, a computer vision product that is in millions of hoees. It wouldn't be hard for google or apple to this but they haven't for some reason.

The problem with microsoft has been that is ran it's research wing separate from the business, while google runs research like a business http://cacm.acm.org/magazines/2012/7/151226-googles-hybrid-a...

A full shift to the realization that machine learning isn't research and should actually be the majority of what everyone does hasn't sunk in it. The question not be "where can we apply machine learning" but instead "where should machine learning not be applied". This type of thinking should occur even down to the individual person (quantified self) http://quantifiedself.com/ .

I don't understand your point. When a problem that needs ML materializes, someone uses it to solve said problem. When someone's working on games, firmware, operating systems and compilers, why would they need to come up with creative ways to use ML when they could be doing higher priority stuff (like actually producing a good game?)

Why is ML superior to (say) model checking, numerical analysis and compilers that everyone needs to be doing it?

Just flip that around. First consider whether you can get data to solve a problem, and if you can get sufficient data, do machine learning. What you basically get is test-driven development with the development part being done automatically.

If you can't get data cheaply enough, only then consider your normal choices. It's easier and less creative to just use the data.

Reply on news.ycombinator.com