HPBigData2014

Raj talking about the difference between #predictiveanalytics vs #prescriptiveanalytics great conversation on #bigdata

Complimenting Google, Yahoo!, and Bing's admarket and userbase -- Raj says that the search market is highly fractured, as search is everywhere, even inside apps, not all search is on any single search engine.

I agree with Furrier when he said that much of search is about user experience -- when people have too much data in front of them, display becomes king when it comes to prioritizing, organizing, and giving context to information.

Searching ephemeral experiential things is a totally different problem than searching for sdtatic info.

So Google can tell you ford instance, how to spell a product's name. It can't tell you how many people are looking at tdhat product online right now.

Both are valuable for different purposes.

Real-time analytics is a very big and important part for DevOps as well (noting Furrier's question on Business Intelligence) as with everything having smarter automation to provide real-time insight can mean quicker reactions to early trends.

@oxidizingangel Facebook retargeting is killing me if I have to look at that damn Sheraton ad again i'll go crazy

AWS (and other global IaaS) can provide precisely dthe same stdack anywhere worldwide, so your software runs exactly the same if you want to move a copy to a new geography.

@furrier The downside of Cloud advertising!

@furrier A problem with that is u have made ur purchase but Facebook is still sending the ad to you. That is ineffective advertising and it annoys the customer.

@furrier There's also a personal aspect to user experience (and business intelligence) especially when users think of websites as people and websites target ads at them -- that relationship means users get frustrated by aggressive or dumb ads.

There's also an element that unlike more passive advertisement methods (TV, radio) the ads are harder to escape/ignore and they're tailored "thoughtfully" by a BI algorithm. Part of that thoughtfulness really may need to judge user reaction.

Another issue with that is sometimes you are not a customer at all. I do lots of searching on things like flash arrays to check product names for articles. Then I get ads for flash arrays.

@oxidizingangel I can't fast fordward throu them certainly, which is what I do with TV ads most of the time. I do ignore them often.

interesting concept re: prescriptive analytics

@dvellante @oxidizingangel it would be great if FB could sense the speed that I pass over the ads as I scroll and learn from that

Colin Mahony see's column store as a need most if not all people will need; but says customers don't want a one size fits all model so Vertica is responding to that need

Mapreduce is pretty much dead; focus has expanded to other areas like Yarn and Haven

@dvellante everyone is frustrated with legacy datawarehouse model - that market is being disrupted

RDBMS is also very expensive compared to Hadoop & the noSQL approach.

@furrier what's the reasoning behind saying map reduce is dead? Sounds like a fluff CEO statement.

@adron totally agree I just asked him that exact question - i'll ask him again go to live.siiconangle.tv to watch now

especially for non-RDBMS data... I've always thought it insane that for decades IT always attempted to shove EVERYTHING in an RDBMS.

@adron that is changing now - i think startups and companies who focus on data not the database will win

@furrier link didn't work.

@adron live.siliconangle.tv

Now the opposite is happening -- companies r putting everything in noSQL, including structured

@adron he just answered your two questoins

@Aron Hall: http://live.silicona...

@furrier ohhh, he's talking about the hadoopi mapreducy thing, not the design pattern.

Colin says "people don't want to structure info before they get value from it." Google moved from MapReduce 5 yrs ago.

@furrier I was about to say he's making fluffy statements, but he said it first. Maybe I like this exec (err, GM). Albeit I'd toss the car analogy... most would kill themselves with a Ferrari.

@furrier anyway, good interview. I'm actually interested to see more of HP's tooling and cloud services become reality and drop to release.

@adron that is a big issue yes some tooling can be challenging

hadoop also much faster ingesting data dthan an RDBMS, which is a big issue today.

Player behavior in free-to-play games is a primary signal as to how much players will pay for a game that they receive for free. Just like any other advertisement market, it's useful to know *when* to offer upgrades for money.

I hate the "let us farm all your e-mail condtacts" growth hacks. Many of those people I have no real contact with. They are former PR contacts from yrs ago.

Peek games talking about the changes at Zynga as example of the massive changes in the industry

@oxidizingangel You need to hit them at the "decision moment" when they are deciding whether to move to tdhe paid version.

Are there any current data models that use player behavior in video games that can help avoid the perception of pay-to-win (a very big turn off) when it comes to free-to-play games but still make money from player needs?

I would rather buy into a game than have them then "sell" you add ons, etc. once you are in. That is a way to make money, but for the player the problem is tdhat the person with a big budget buys a win.

For Inan: What kind of metrics do you generally want to pull on player behavior, do you have an example of what data models you use to judge player behavior when it comes to who will pay money and how non-paying players help raise revenue?

Kim Kardashian is apparently making a fortune doing that with her game.

Very interesting how they use data to balance the game to maximize player satisfaction -- getting the right win rate -- without disrupting players.

World of Tanks also received a Guinness Book of World Records record for most concurrent users on one server. Wargaming.net is pretty amazing when it comes to IT capability and sheer popularity.

So I can make money on all my tweets and posts? I like it!

What is the value of that data? Is it worth paying for?

Data is an asset without question. How to share the $ split when using it generates revenue is a sticky challenge

@BertLatamore -> Google makes more money from your data, than it costs to deliver the services that gather it.

this will be the biggest issues in next decade I love this point

I think that personal brokers will be in play and people will give up (authenticate) for value; FB for instance is making millions from data for the right to use the software #moralhazzard ?

no, I don't think so. Consumer gets value from services delivered in exchange for his/her data. If no value being delivered, consumer should stop handing over his/her data

the data mafia will rule

There's too much data already collected that's not "owned" by the consumer for them to start getting paid at some point. Brokers may move significatly down the stack, as sensors become cheaper.

imo the consumer gets "value" in exchange for data/personal information - ideally opt-in but without tighter privacy laws...