What's behind the Curtain?

What's behind the Curtain? I was listening to a podcast today which made me think. "Aleks" Krotoski contributed a story to the Tech Weekly podcast. For those who would like to hear it you can find it at http://www.guardian.co.uk/technology/audio/2013/jun/27/podcast-tech-weekly-mendeley-victor-henning and the story starts at about 18:17 from the start, or 9:15 from the end. What is the difference between Google Now which purports to know what we want, sometimes before we ourselves know what we want, and the Helpline which is able to answer such questions as "I have 3 weeks in Europe. I've never been before - where so I need to go?" or "How much will I spend in 5 days in X?". Are we like an app which knows all? Have a listen to the portion of the podcast and please tell me what you think of her piece, and this question. What is behind the Curtain?

Posted by Rose
NYC
922 posts

I haven't had time to listen to the podcast but Agnes' replies really ring true for me as well. I can't stand being marketed to in any form. It's why I don't have a TV hooked up because I don't want the constant commercialism and advertising invading the quiet space and peace I value so highly in my home. And I have ad-blocking technology on my browser so I surf the Web almost completely free of invasive adverts. I also don't like it when technology tries to tell me what it thinks I would either want or 'need' or should think or do based on my past search behavior on the Web or anything related to it. I have very broad, varied, and eclectic interests, and I dip into them randomly. No algorithm will ever be able to define me or persuade me. I've never understood the proliferation of billboards along highways. I don't think I've ever made a purchase decision based in what I glanced at for a few seconds on a highway billboard, and I rarely make purchase decisions based on advertising or any type of marketing. No person, group, or technology will ever be able to pin me down or label me. I dance completely to my own drum, and the tune changes constantly.

Posted by Ms. Jo
Frankfurt, Germany
4759 posts

Which Curtain? #1, #2, or #3? I'll take curtain #3 for 10,000 please and can I buy a vowel too? Nice Monday post there, Nigel.

Posted by Kevin
near Ringwood, Hampshire, UK
521 posts

What is the difference between Google Now which purports to know what we want, sometimes before we ourselves know what we want, and the Helpline which is able to answer such questions as "I have 3 weeks in Europe. I've never been before - where so I need to go?" or "How much will I spend in 5 days in X?" One difference is that Google Now knows something about your interests, or at least what is in your search history, while the Helpline usually has very little information to go on. Specific questions, about itineraries for example, tend to be about the same fairly limited list of places and that's because they are just a reflection of the guidebooks that everyone buys. Google Now suggests the same things all the time because it doesn't know what else you might like, the Helpline does it because so many people ask the same questions. On the other hand, I think the strength of the Helpline is that it doesn't just advise people about London-Bath-Cotswolds-York-Venice-Florence-Cinque Terre, Euros at an ATM, SIM card from Carphone Warehouse and how to guard against pickpockets (although of course it does all of that very well). If you look beyond that you find a genuine depth of experience that people are glad to share. Google Now will never be able to do do that.

Posted by Agnes
Alexandria, VA, USA
616 posts

I like this Helpline primarily because it's basically staffed by (mostly) kind, generous human volunteers who "man the threads" and post answers without any reward except "thanks" and paying it forward. BUT at the same time, from a technological standpoint, this helpline is a rather inefficient format for aggregating and sorting a "body of knowledge", if you will (the saving grace is people respond very quickly). Think about it...the same questions get posted over and over again, it's obviously cumbersome to search old threads (poster doesn't want to "work" at searching vs just posting another same question adding to the clutter - i.e. my favorite is "how do I exchange Dollars for Euros?" ad nauseum. Then the questions just get buried under a long stack of questions and rise to the top via popularity contest and interest of posters (or there's a food fight going on between posters). Then there are the 2012 or older postings that appear out of nowhere (that I've gotten caught answering, sigh)...is it just to create some dynamism or is it a weird fluke? If Google algorithms were applied to this Helpline, I bet the system can sort through the giant mound, remove all duplicates, aggregate the stuff into a hierarchy of responses, and develop a profile of each poster and respondent based on their history (especially the folks who have > 1000 postings or some threshold to make decent assumptions). When someone would start to type a question, Google can probably autofill or suggest that it's been asked before. Less clutter = good. Google mimics a human's learning process - it needs constant feedback and correction to "learn" and develop heuristics. So it makes sense that it feeds on information. The more it has, the more fine tuned it can be. (to be continued....)

Posted by Agnes
Alexandria, VA, USA
616 posts

What google does is not at all different than how people think here...for example, if I go through and read all of say, "Suzy from CA" posts, I bet I can pretty much develop at least a soft understanding of Suzy, her philosophy and biases, and what she's likely to recommend..due to repetition of postings, it's easy for a human to make certain assumptions. If google went through this database of helpline posts, it would do the same thing except with algorithms, which would be consistently tested with each new post. Humans couldn't handle all that info on all people at once, that's the benefit of technology. Humanistically, I totally reject predictive technologies as a "normative good" not to be questioned. I don't necessarily want a similar experience, a similar book, a similar restaurant, or a similar hotel or trip than the one I had before. I don't want to be a creature of comfort, narrow tastes, and I like to stumble on things I had no interest in whatsover. So I don't welcome the intrusion of some gadget nudging me to do certain things at each decision point - I'd end up telling it to shut up. On the other hand, if marketers can figure out what you really value (through constant iterative feedback from you), they'd probably waste less of your time peddling things you have 0 interest in, which seems like a win, win. I personally just want to minimize the level of marketing I have to put up with each day - but to each his own. I want to control whether to opt in and to opt out - so I wouldn't tie myself exclusively to Google Now. There is a good point made by the video - in order for these technologies to be useful, they need a robust database behind them. Except for throwing out new ideas to try, Helpline posters can't possibly know much about the motivations of various posters unless they've had lots of prior interactions with them.

Posted by Jim
Dallas, Texas, USA
495 posts

Agnes as well as others, I think I know what you might really know, and what I think you would like, but not sure if I should let you know just yet. Give me a few minutes to BING-it, and then I'll get right back to you! Nigel, wonderful topic! I agree with Agnes and Rose.

Posted by Agnes
Alexandria, VA, USA
616 posts

I should clarify something I said before, although my answers have been pretty long winded. I actually don't mind getting marketed to IF it's something I value. For example, if google knows that I love to go to Florida in the winter, then bring on the cheap flight deals so I can jump on them! It saves me time and money from combing through fares all the time. In general, it's more efficient that some service provider who knows what I want actually responds with an offer that I value. But it's hard for technology to do that without having lots and lots of data points - in which case it would keep "trying" to sell you "a", "b", and "c" (along with what I really want) just to test my boundaries. I think that's when people get irked. Not that different from human relationships...ha! The devil's bargain is you have to give a lot of personal info up and then you'll get more accurate predictions (hopefully)...assuming that's what you want to begin with. Now to loop back to this Traveline, I read it for entertainment value and to connect with nice folks (and yes, get my Qs answered) but I don't pretend that anyone really knows someone else's preferences from a 1-2 sentence question (e.g. "I'll be in Europe for 2 weeks. Where should I go?".....um, how do I know?)

Posted by Keith
United Kingdom
680 posts

Nigel - If you want to know the answer before you think you've asked the question, surely you should ring up GCHQ. They will have followed your internet planning, read your draft questions and had time to look up a response. Perfect service.

Posted by Agnes
Alexandria, VA, USA
616 posts

Nigel, so what are your own thoughts after listening to the podcast?

Posted by Rose
NYC
922 posts

We can only hope for further development of a true 'semantic Web' based on the brilliant, ground-breaking research of Sir Tim Berners-Lee.

Posted by Nigel
East Midlands, England
8741 posts

No, I don't think that Helpline is like an app that knows all about the people who ask questions here, especially if the questions are particularly vague. I think that the podcast excerpt from Aleks also shows that even the Google Now algorithm doesn't really have the ability to know what people want, even if it sometimes seems so. I may have a reputation around here for asking a lot of questions of posters; because it is so critical to be able to provide the answer a poster needs rather what the answerer either likes themselves or guesses that everybody would want. I'd rather that posters shared themselves from the beginning so that sculpted and tailored answers could be accurately crafted. I guess that Google would like more data too.