I think readers know that on balance, I’m a fan of Google. I recently switched to the Nexus 4 (more coming on that front as I settle into really using it). I believe the company has a stronger core philosophy than many of its rivals. Overall, given that it’s nearly impossible to avoid putting your data into someone’s cloud, I believe that Google is probably the best choice for any number of reasons.
But that doesn’t mean I won’t criticize the company. And every year about this time, I end up doing just that.
Because the annual Google Zeitgeist came out this week, and I’ve spent a bit of time digging into it. And once again, I’m pretty disappointed.
In the past I’ve criticized Google for failing to ask interesting questions of the massive amount of data it collects on search patterns each year. Once again, this lament applies. I honestly do not care what top ten TV Shows, Sports Stars, Songs, or even People we collectively care about, because there is *never* a surprise in those results.
But Google knows so much more….and could really tease out some insights if it cared to. Imagine if Google took its massive search query database and worked with some of the leaders in the open data movement to mine true insights? Sure, Google would have to be careful about how it released the data, but the output would be extraordinary, I’d warrant.
Instead, we find out that Gangnam Style was a big deal this year. No shit!?
But it gets worse. Not only is Zeitgeist rife with pop culture fluff, as you drill down into it by country, eager perhaps to find something interesting, it turns out Google has chosen to eliminate certain potentially sensitive categories altogether.
For the US and most other countries, for example, there is a “What is….” category, which shows the top search queries that start with “What is…” For the US, the answers are
- What is SOPA
- What is Scientology
- What is KONY
- What is Yolo
- What is Instagram
- What is Pinterest
- What is Lent
- What is Obamacare
- What is iCloud
- What is Planking
But is there a “What Is…” for Saudi Arabia? Nope. China? Uh-uh. The United Arab Emirates? No sir. Egypt? Move along.
Hmmm.
Oddly, Google did provide “What is…” was for Singapore, where people living under that “benign dictatorship” were interested in the same things as the US – “What is SOPA”, “What is Scientology” and, for politicians, who is “Mitt Romney.”
For the US only, you can drill down into all manners of other categories past the main page, including News, Science, Tech, Humanities, and Cities. Those are pretty interesting categories, but Google only provides them for the US, which is a shame.
Furthermore, I find it interesting that Google, with all of its translation technology, does not have a translation button on the results pages for countries where the majority of the searches are in languages other than English. This is most likely due to political sensitivities, because if you run some of the results through Google Translate (do you believe I had to do that?!), you get some stuff that I am sure does not please the regimes of countries like China, Saudi Arabia, and the UAE.
For example, here are some of the top searches for Saudi Arabia, translated (roughly I am sure) by the Google Translate service:
Student outcomes Arab Idol insurance Ramadan Series 2012 Mohamed Morsi explosion Riyadh Burma Free Syrian Army Shura Council tornado Sandy
But again, you aren’t going to get much more insight into what Saudi folks are *really* thinking about, because Google failed to ask the interesting questions, like those it has in the “News” section of the US Zeitgeist. I’d sure be interested in “Political Gaffes,” “Election Issues,” and “News Sources,” in Saudi Arabia, China, or the UAE.
In fact, for Saudi Arabia, Google has ommitted the “Top News Searches” box that is on several of the other country pages (even Egypt). Instead, the topics for Saudi Arabia (besides trending searches and people) focus on sports and entertainment stars, fashion designers, TV shows, and the like. Deep, Google. Thanks.
Now, the datasets are different for each country, and it may be that Google simply didn’t have enough trending data to surface interesting political insights for these controversial countries.
Somehow, though, I don’t buy that. This set of lists feels extremely human vetted – I’m guessing an awful lot of hand wringing went into chosing what to show and what might prove problematic to Google’s best interests were it to see the light of day.
If that is the case, I urge the company to have more courage. I bet if Google open sourced its query data sets (eliminating any chance of PII getting out, of course), I bet academics, data scientists, and just plain interested folks would let loose an explosion of insight. Pop up the rainbird of data, Google, and let the ecosystem flourish. We’d all be the richer for it.
First, giving third party access to its data will make users uncomfortable, I know no PII, but most users don’t understand these things, it’d just give google’s rivals more ammunition to undermine users trust for google. second, if they open the data to the public and even average folks can draw some insights, it’d destroy many market research firms.
I think this can be managed, though yes, there are risks.
You are so irrelevant now.
John,
As always, enjoy reading your search blog. For another perspective, here’s a non-vetted set from our Hitwise U.S. data-set (10mm U.S.) for last week:
1. What is my IP
2. What is Gangnam Style
3. What is the fiscal cliff
4. What is gluten
5. What is the Illuminati
6. What is a thesis statement
7. What is Obamacare
8. What is love
9. What is squab
10. What is right to work.
This covers all major, secondary and tertiary search engines (58 in total). Looking through the 38,000+ “what is” queries that we have access to, the top trends appear to be news/media driven informational requests, pop culture and simple tech questions.
Thanks for the new data!
But Google knows so much more….and could really tease out some insights if it cared to. Imagine if Google took its massive search query database and worked with some of the leaders in the open data movement to mine true insights?
I haven’t commented much in the past year or two, but starting in about 2005, that was one of the common themes running throughout my discussions here in your comments section. Google DNA is not about algorithms that help the average user gain insights. Google DNA is about algorithms that give the average user easy (aka unsurprising) answers.
Think about it. Take your objective criticism of Google Zeitgeist, and think about Google Search, the flagship product itself, from that same perspective. When you type in a word, two words, three words.. what do you expect to get back, as a search engine result? Are you going to get back “insight”? Or are you going to get back expected, known, popular, unsurprising, common results?
The latter. That’s just the way Google works. You’ll get the latter.
Now, don’t get me wrong: It’s an amazing technical feat that Google has been able to automate the mass scale algorithmic process of returning the “zeitgeist” of search results. And a lot of time, this is exactly what consumers want.
But what *I* want is insight. When I am searching.. I mean really searching.. I don’t want the popular result. I don’t want the common result. I already know that result. What I am looking for is what I don’t know. I’m looking for the surprising. For the unexpected. For insight.
Frankly, I’m beginning to really tire of web search in general, because even though I’ve been ask for insightful search results for over a decade (I first started talking about it, privately, in professional circles in 1999.. when Google was still building its servers out of legos), nothing has changed. They’re still the same old boring, common search algorithms. Great if I want to look up the nearest Chipotle. But useless otherwise.
JG your critique of Google and search have always been insightful. But I think in fact Google is really leaning toward insight as a goal, at least from my discussions with the team there over the past year. See the Wired piece I just RT’d for some (but not all) context…
Hmm. I just now read the article, and I’m still chewing it over in my mind. I’m sure I’ll have more to say later, but it will feels like 90% of the examples given in the article are more of the same “give me the expected, known, popular, common, zeitgeistian answer”.
What they’re doing is making that answer more accurate: Giving you the sugar shack in Massachusetts if your phone says that you’re closer to that one, not to the sugar shack strip club a thousand miles away. That’s great. There is tons of utility to that. But that’s not what I’m talking about, when I’m talking insight. It’s just making the computer better at giving the expected answer. If I’m three miles away from the sugar shack in Massachusetts, then I already expect that the one with maple candy, not Candy, to be my answer. I gain no new insight when Google smartly uses my location to answer that query. I just get my expected answer, easier.
The bit about the knowledge graph did get a little closer to what I’m after, but even there, the knowledge graph is an attempt to label everything in the world with some sort of semantic markup, and then link together the most common relationships between those labels, correct? To me, the whole reason I’m searching is because I’m trying to find fact relationships that *aren’t* commonly known.
I may have an incomplete understanding of what Google means by it, but to me it sounds like the knowledge graph is mostly about figuring out that hot dog buns are linked to hot dogs. True. But expected. I want insightful. I want to discover that hot dog buns are related to the French-Indian war. And when I find out why, I don’t expect that the answer will come in the form of a single web page. There might be three different hops to connect these things, and if I don’t find all three, then the connection is never made. Finding only two of them isn’t just a partial result. It is a non-result.
However, I am encouraged by what it says at the very end, about the future of search being more about verbs than nouns. That resonates with me.
It takes steps, and also, you have to have some way to indicate you are in “insight mode” or at the very least, have a mode that is available to indicate that. Inferring that you are looking for insight is damn hard human to human, unless you’re on a psychologist’s couch…
Yes, absolutely, you have to be able to indicate that you’re in “insight mode”. You wouldn’t want it always on. But the fact that you need the user to indicate the mode that they’re in is one of the reasons I think Google will never do it. It’s against their DNA to “clutter” the interface with user choices. Steps are indeed needed, and their unwillingness, for over ten years now, to even take that first step is why I was saying earlier than I really am beginning to tire of web search. It really should not be called web “search”, because they do not offer the tools to actually let me dig in and search, such as this insight mode button. Rather, it should be called “web easy answering” or “web popular fact regurgitation”. Hrmph.
You are so damn right, sir.