Search relevance loosely refers to whether a query returns relevant results. There are different ways of measuring this, none of which is really bulletproof.
from that query how does one know user's true intent? Is he looking for a wikipedia article? is he looking to hook up? is he looking for porn? is he referring to the gender? is he just bored and trolling?
since google's reach is part of our lives, wouldn't they have a very good idea based on your previous history, what you clicked on, what search results you clicked on, what you typed in google chrome.
couldn't microsoft have pulled off the same thing?
You don't know for sure what a user means by a query. You can make informed guesses. Typical tricks include looking at what users in the same area have searched for, looking at previous queries (turns out a lot of searches are repeated), looking for "replaceable results" (results that appear on multiple pages), and so on.
And no MS could not have pulled off the same thing. The data MS has about you is not even close to the data GOogle has on you.
> The data MS has about you is not even close to the data GOogle has on you.
Also, users have adapted and are trained to work with Google nowdays, they memorize how to find particular results on Google (even though Google puts a lot of effort into breaking features like "link:...").
At Google there is an internal metric that always shows Google search quality leading Bing's. That's fine, but I always wondered if Bing has the same internal search quality metrics and they are always ahead?
Tonight I typed Eagles into Google and it showed me all the information I needed about the game that starts at 8:10 pm. That is search relevance. I've noticed Bing doing a lot of this type of thing for the past year or two but Google is definitely the master of it.
Last night I asked Google how much a driving theory lesson costs. It spoke the answer back to me that it had extracted using NLP from the official website. That is the power of their infrastructure.
Since Google knows a lot about me (for eg) by parsing my emails (I use gmail) and knows about the sites I visit too (Chrome and thru Adsense on 3rd websites), Google is in a much better position to provide me with more relevant results(and ads) than Bing, since I don't use much Microsoft's online services.
Google has a lot of ways of knowing what I want to find, which definitely has disadvantages; I get presented with the same ads & results even if I try to find other stuff. I get presented with the same article or ad in different searches numerous times to the point I go on to anonymous browsing / ddg.
The good thing; when just searched for something about UIView and later type something "populating a table", it'll give me SO articles (and such) about UITableView which is what I meant but forgot to type add ios/or uitable.
The bad thing; when i'm doing research, it'll give me the same stuff over and over again and I have to page quite far and add / remove query info / play with search tools to get to different things which might not be the most popular or might not be what Google thinks I want.
The awful thing; the ads... I use ad blockers, but when I get ads (gmail / some sites), they are always the same. For Facebook I just think their algorithm is broken or maybe they have none (I only get ads about dating, cheating, stocks all of which I have no interest in nor ever shown), but for Google they show things I am interested in but I already know all of them, have been there, knew most of them before those ads and they cannot possibly sell me anything. If I wanted something from them I would go there without those ads, so why are they targeting me like that? Show me ads I don't know and if I can use it, I would click.
I'm technical, but I hear the same annoyance from others; a friend of mine rents out houses; the search results when he wants to research the competition are usually almost only his own sites (presumably because he visits them a lot?) and ads for his own properties. He is very non technical, so explaining this all to him is not really sinking in.
This is a great example of search result subjectivity. I don't care about football. I'm annoyed when it rudely intrudes my consciousness. If I type "eagles" into a search engine, I'm looking for the band. The football team has no relevance to me, even as millions of other people tonight are probably looking for exactly what Google showed you.
Yep, and that's why it's important for the search engine to be able to collect information about you! It's not just for commercial ends like ad targeting.
I find this title to be misleading. It implies the post will demonstrate what viable search engine competition would look like, but just explains the problems that Bing is having with Google. Are we to assume that Bing v. Google is viable search engine competition?
While Microsoft's problems are real, this offers little in the way of solutions. That said; interesting read.
Hmm, I didn't really say that I had the solution. If I did I'd start my own company. :) I said that I would outline the state of making search engines competitive. Here I basically focus the post around areas that are pain points for competition, which I think is pretty relevant to the writeup I set out to make.
Do you feel something in particular sucks about the article? I'm happy to change it!
That said, I use Bing as a comparison point because it's really the only other point that's comparable in terms of, like, infrastructure and stuff. It's contextualizing more than anything. I might have removed all the references to "Bing", but I don't think it would add clarity.
Hahaha, well, my bias is certainly not minimal! I work at Microsoft!
But I see your point.
In fairness I've tried to be honest about our weaknesses and not shy away from the things that are bad. I think I've been reasonably successful, but of course I'd love to hear from you if you think I'm giving MS too much credit.
Given the article is explicitly about things he's learned about the process of competing with Google through working for Microsoft, the fact that he works for Microsoft is precisely why he's credible.
What would you prefer? Somebody with no relevant experience talking about the subject? I'm sorry, I really don't understand how actually knowing what he's talking about 'marginalises his credibility'.
Short and very informative, though I wondered why patents are no concern for a competitor of Google (they have various broad and trivial patents on using historical data, user feedback, moderation for better search results).
As the author wrote, aiming at feature parity with Google is probably a bad idea. DDG seems to do this right: they have some better features and one distinct advantage over Google with their stance on privacy. Perhaps Google will reach a certain creepiness factor with all the collected data they're factoring into search results (e.g. I have this suspicion that content from gmail is used too) and at some point it might work against them even for the average user (or am I just biased?).
One "feature" that Google has failed to do properly is custom "site search", i.e. providing a search interface for arbitrary websites. Perhaps this is another opportunity for a new competitor or even Bing (though their "site search" was shut down in 2011 apparently), there is plenty of demand (and lots of crappy custom solutions).
I wrote an opinionated blog post exactly about that saying[1]:
"When each site understands its own data better than Google, its internal search results will surpass Google’s. Google will no doubt continue to provide better global results, but the two-tiered search would decentralize efforts to improve algorithms."
I see it as one of the Achilles' heel of Google. Even if what they are doing seem impossible to copy.
Currently, it is very easy for websites to provide better search results (but still not good) compared to Google and other search engines like DDG due to the limited crawling, except perhaps for the top few 1000 websites with small enough number of pages/URLs so that at least Google can keep them fully indexed.
This is also where dedicated "site search" providers can beat these search engines easily: they can offer a full crawl and depending on the size of the resulting index, they can bill the website owner (e.g. by storage used in GB or number of URLs).
Site search is an entirely different problem than searching the web, in fact, really good site search for 100-1000, 10,000-100,000 and 1,000,000+ are entirely different problems.
Because Google usually doesn't use its patents aggressively. They will use them defensively, though, or even as a counter-attack. That's why I think it's incredibly stupid of Microsoft to be attacking Google with search patents through Rockstar. If they piss off Google enough to start using their search engine patent trove against Bing, it may be game over for Bing. Microsoft should already consider itself very lucky that Google isn't as spiteful and vengeful as Apple, or Google would've used them since day one of them going after Android OEMs.
That being said, I think the only solution to beating Google is developing something disruptive, and I mean that in a very classical and traditional sense (not the way disruptive is used to mean anything these days). It means you have to build a type of a search engine (or search solution) that Google not only has no interest in pursuing (due to conflicts of interest/opposite business incentives), but probably can't pursue, or not without employing dramatic shifts in how they do search (which again, makes it even less likely for Google to pursue and try to beat this new "disruptor").
And no, don't look at Bing to provide that. They're a "direct competitor" to Google, and they will always be, because they want to be "in the exact same business" Google is, and therefore, I don't think they'll ever radically change the way they build a search engine compared to Google.
So what can such a disruptive solution be? I can't say for sure, because it probably needs more factors/benefits on its side than just one, against Google. Social "may" be a way to disrupt Google, although I think Google is already rapidly adapting to this attack vector.
I think another could be "extreme-privacy", which Google wouldn't pursue anytime soon without really rethinking how its search business works.
Disruption also means attacking incumbent companies with much fewer resources through innovative business models and cost structures. Do something that would make Google's billions of investment in infrastructure irrelevant. A P2P search engine like YaCy or Faroo seems to be part of that solution. Perhaps what can/will kill Google is a crowdsourced super-search engine, that ends up surpassing even Google in most capabilities, because Google would be just one company, fighting against the whole resources of the world.
As in classic disruption, you also have to consider this new type of search engine will be much poorer than Google in some competition factors (at least initially). I think that is the "relevance" factor. As the op mentions, it's going to be very hard for anyone or anything to beat Google in relevance. They have too much of a headstart. But perhaps there will be such a thing as "good enough" in search, too. Maybe we're not there yet, but perhaps we will be in 5 years, or 10 years. Until a few years ago we kept wanting faster and faster Intel processors. Now we think mobile ARM processors are more than good enough for most of our daily activities.
A crowdsourced P2P search engine powered by billions of people could perhaps reach that point, too, while offering some of that "extreme privacy" and other benefits that Google can't offer.
Still, how do you get that first "10 percent" of users, that end up changing the paradigm, to start using a service that's much less relevant than Google today? You do it in the same way other disruptive technologies win out against incumbents. You promote the other benefits, that Google can't and won't match - benefits that will be important to those "first 10 percent", until you reach critical mass, and then you should be fast approaching that "good enough" relevance, too, and get even more users after that.
Voice search with punctuation. I would use Bing if it provided this. As for now I use Bing simply to make sure I'm not loading cached pages on bad connections.
> Because Google usually doesn't use its patents aggressively.
And that's probably because Google is incredibly weak when it comes to patents. This is kind of well-known in IP licensing circles, but it's also evident in their desperation to acquire patents (Motorola, Novell, Nortel... and other patent holding companies that nobody's never heard of).
It may not be in their nature to attack, but even if they wanted to, they are much more vulnerable to a counter-attack.
> I wondered why patents are no concern for a competitor of Google (they have various broad and trivial patents on using historical data, user feedback, moderation for better search results).
How exactly are server-side patents enforced? When the hardware on which the infringement is occurring isn't exposed to the public, how can anyone verify its presence? I know that for some (many?) of Google's search patents, there might be an indication on the client side if infringement were occurring, but I imagine that there are also many patents for which this isn't the case.
You don't need convincing proof, just reasonable suspicion to file a lawsuit. (There's something called Rule 11 to discourage frivolous lawsuits, but it's a pretty low bar.) You can then determine actual infringement during discovery.
This has some problems however:
1) As you can imagine discovery may not turn up any actual infringement, making it all a huge, expensive waste of time.
2) The unfortunate reality of patent lawsuits is that if you can't prove infringement just by looking at something, you might as well have already lost. Anything that requires expert witnesses to provide input on often comes down to which sides' witness the jury finds more likable, and that's pretty much a roll of the dice.
If Bing were to overtake Google and become the preferred search engine, would anything really be much different?
I'm no search engine expert but I'm pretty comfortable assuming that Bing would have(and does have)had the same issues with rap genius as Google did. The only real difference is that as the market leader Google had to publicly react to rap genius 'gaming' of their system.
I think what this shows is that yes, this is precisely a product problem because if Bing were at the top we would be having this exact same discussion only with reversed roles.
Correct me if I'm wrong but Google vs Bing is not a lot different than Coke vs Pepsi.
Until we have someone come along with a search engine which is actually a better product, market share will only really be a function of effective marketing.
That said, I think your article does a good job outlining some of the barriers that any potential competitor in this arena should be aware of.
The common "problem" that Bing and Google could both potentially be upset with RapGenius is the there is no accountability to the site owners. The search engine's job is to serve results that people want to see. In general, you can absolutely count on them to make decisions that they think will benefit their total relevance in aggregate, even if it hurts a particular subset of queries.
But that's not really so much of a product problem as a problem with the problem itself. In other words, it's not really clear that a better product will fix that, because the incentives will probably always point to doing this rather than not.
> If Bing were to overtake Google and become the preferred search engine, would anything really be much different?
No one wants that (except Microsoft).
What we want is for Bing or something else to bring more balance to the search market. Today Google is so dominant that people SEO specifically for it, and Google decides what is and isn't ok to do, and can then decimate site's traffic based on that (as we saw with Rockstar in a recent example).
But if there were several successful search engines with no single dominant winner, things would be better. SEO would matter less because you would need to optimize for multiple targets, so you would get away with less tricks (or you need to work much harder to trick everyone, again with the result of less trickery). And no single company could decide the fate of every other company that needs to get traffic through web searches, which is what we have now.
"If Bing were to overtake Google and become the preferred search engine, would anything really be much different?"
Maybe. Google sells site ads; Microsoft sells Windows and Office licenses; DuckDuckGo sells search ads; Apple sells hardware. Following the money, you realize that Google's goal is to produce search results just good enough to maintain its brand, while driving as much traffic as possible to sites that display its ads. Always ask yourself how your interests differ from your search engine's.
Imagine if Microsoft begins saving users' information using its operating system. As result they got more information on you even than Google or anyone else.
I enjoy hating on Microsoft as much as the next guy, but as far as I'm concerned Bing is basically as good as Google.
The problem for Microsoft is the same problem that faced Apple. Being "as good as" isn't enough to get people to switch. You have to be 10 times better.
Back when Google launched, it was 10x better than Yahoo. Bing seems to have quickly gotten to 1x, but hasn't moved much further.
"The problem for Microsoft is the same problem that faced Apple."
Pretty much: Apple was faced with a competitor who owned a vast majority of the PC market. They never took over the majority, but they managed to earn a profit from a minority of people who were willing to pay a premium for computers that worked better than the crapware offered by most Windows OEMs. Windows was cheap, and just good enough to write TPS reports.
Google now offers similar crapware: heavily-gamed and -"personalized" search results slathered with ads and just good enough to be better than most alternatives. There's plenty of room for a company to offer something better: paid subscription search, less-complete search with fewer ads and more accuracy, or something else (I'm not a "10x founder," so I can't think of these things.).
I still regularly make searches in duckduckgo (which gets results from Bing) and then find that I have to switch Google to actually find what I'm looking for. The worst thing about this is that Google's results don't seem to be as good as they used to be, and there isn't really any other search engine that matches it yet that I've seen.
Hmmm but DDG will never be quite as good as Bing or Google simply because DDG doesn't keep any histories or track any other activity. Really good search engines get quality and relevance from all those things that DDG purports NOT to track. It's partly a machine learning problem that I just am very very skeptical DDG can get around.
Exactly. Personally, only a very small fraction of my searches are 'random', about something I've never browsed about before. However, most of my searches are either very predictable (related to stuff I browse, search every day). Or I'm searching for something I've browsed already just didn't bookmark or can't bother to find because I know google will find it.
I don't think this is true unless you are truly locked into your own tiny little world. All your search history really does is help Google serve up ads for you. If that's an improvement, so be it.
Many programming language related terms, movie titles, cities/postal codes, ... All those things can be judged better when the search engine has prior knowledge about you. You don't have to be "truly locked into your own tiny little world" to have previous search be a valuable predictor of what kind of results you expect.
Google keeps track of what sources you're more frequently clicking on and weights results subtly but noticeably. If you look up things on Wikipedia a lot, Wikipedia pages will be ranked higher. If you're more likely to click over to the BBC than to Fox when you read a news source, the BBC will be higher and Fox will be lower.
I choose to stick with google because given the choice, I'd rather see them succeed than microsoft.
We're in the predictable phase where everyone hates google now because they're no longer the next big thing, but I'm not a hipster and I don't think microsoft is now the underdog so I don't get the google animosity and I don't get the sudden warmth towards microsoft.
Quite frankly I fucking hate windows, and I wish it would just go away entirely. To that end I avoid giving microsoft money at every opportunity.
I don't think Bing is as good as Google, no other search engine in the world can compete with Google, at least when it comes to content in English. Google can almost read my mind, I can't say the same about competing search engines.
There is an interesting dichotomy here. Google does read your mind, in the sense that it uses available information from both your previous searching history and your existing interests in gmail, G+, and basically any web site with a +1 button. So when you search for something it takes all the things that it knows about, prioritizes by the things it knows you are interested in (this is the mind reading part), and then picks the best result to show you. Voila, great result.
If the last web site you visited was for the Audubon Society and the next thing you search for is Cardinals you will get a page about birds on the first result, but if you had just recently looked for tickets to a ball game on Ticketmaster you will get the baseball team as your first result.
One of the more fascinating things that DDG's "Dontbubbleme" campaign did, was expose just how big a swing in the results your 'meta data' has. People who are running Chrome and are "logged into Google" get very different results than people who make a query with no context following them around.
In the latter case everyone has to resort to giving you the pages that "most" people clicked on from your geographic region [1].
Bing and Google have pretty much exact equivalence if you remove tracking and geo context. (IP blind / cookie blind / search).
[1] For even more fun, set up proxy servers in various data centers around the country and do searches with no cookies or context, and proxied to different geographies. Very enlightening.
I use Chromium (logged in) as my main browser but if I try this queries in Firefox with a clean profile I get the expected results. So no Bing and Google are not equivalent I get much better results with Google. Next semester when I return to school I'll check my queries from the school network to see if Google gives me the expected result from there. Either way I like Google more than Microsoft and I rather switch to Yandex than to Bing if it ever comes to that.
Here's a feature, let me add weight to my search results. I want to give preference to stackoverflow, HN, reddit, etc. I don't want to limit my search to only those sites, but I'd like to push them a little closer to the top. Also, let me setup a blacklist of sites to exclude. I don't want to see expertexchange... ever. I should be able to share my custom search setup with others, kind of like how AdBlock has subscriptions, I could select popular setups, or build my own.
And, what if I could select my settings in a dropdown when searching. So, maybe the people at HN setup a whitelist of 100 of the best sites for programming help. I add this list to my account. Now, anytime I want to search for a programming question, I type in the query, select my custom search type from the dropdown, 'programming questions', and then click search. I'll only get results from those 100 sites.
Google will never implement this for a variety of reasons, so it's room for competition.
I am not 100% sure but I believe that Google has some of this ground covered. If I do a search for "Java" (for example, not sure this particular term will produce distinct results) while logged into my Google account, I will get different results than if my non-programming friends searched for "Java." Again, not totally certain that this is the case but I have seen some evidence before that search results are customized based on past behavior and other models of my preferences.
Actually Google did implement something like this: you could "personalize" your search by moving results up and down the page. They discontinued it some time (just a few months?) later... I'm guessing because it created more noise than signal. This was some years ago.
Yeah, it was SearchWiki, launched in 2008, discontinued sometime in early 2011 IIRC. About the only thing people used it for was to kill expertsexchange.com, and that problem took care of itself when StackOverflow got popular.
I have used Bing for years as my primary search engine simply because of Bing rewards. MS essentially pays me to do search (I get about $10 in Amazon gift cards a month from Bing Rewards Gold). At first I found myself first searching on Bing and then going back to Google. Over time however as I have made tens of thousands of searchs, Bing pretty much nails what I want everytime. It took like a year but the gap is pretty much closed. Now I feel like Google's are a little worse because they have way less data and Bing has way more.
What if you made a search engine that only works on specific sites? For the most part users don't want to search the entire web for every small blog in existence, they want to go to a relatively small number of specific sites.
For example, reddit has a terrible search engine that users often complain about and comments aren't indexed at all. But searching with google often gives subpar results and they don't take into account things like upvotes at all. (I also think sites like reddit are punished by Google's algorithm pretty harshly.)
This might make the problem smaller and more tractable and a place where you could get an advantage over Google.
Yep, an effective general purpose search engine is really a collection of effective vertical (subject-intelligent) and horizontal (task-specialized) search engines.
The truth behind IBM Watson is that it's based on a methodology that makes it possible to put huge teams working on different parts of the problem without tripping on each other's feet.
How about distributing the indexing to browser plugins? Apart from delegating the work and displacing crawlers, it would also help with relevance, currency, and avoiding link farms.
points 1, 4, 6 from the article could have been written in 1998 as well.
how can 2 guys in a garage wipe out AltaVista, Lykos, and whoever else was doing search back then (with lots of smart people, data, resources, state-of-the-art software and user adoption to compete against)?
What is missing is a serious discussion on search quality (which made Google Google).
If Larry and Sergey started today, would they be able to compete against Google by providing better search results,
or has search quality reached a "global maximum" of sorts?
AV made a substantial unforced error at exactly the wrong time which resulted in a huge drop in search quality. It wasn't as much a matter of better algorithms as one might think.
There's also a pretty big difference in scale. The AV team working on search in 2000 wasn't all that big.
Personally, I'm amazed that they let him blog stuff in this detail about MS's search business practices. Not speaking for the org or about the official stance, but I would not consider doing so as a member of SI at Google. And I always had the intuition that MS was even more restrictive about such things. Maybe not?
I didn't say anything that isn't completely public knowledge. And let's not forget that MS once employed Robert Scoble. So we can't be all bad.
EDIT: and let's be realistic. The papers on search that come out of MSR expose way more about our infrastructure than the paltry sum presented in this post.
Insiders can often link bits of public knowledge in a way that effectively discloses confidential information. Every company has a different threshold for where this is, so maybe you're on the safe side of that for your org. BTW, I'm not trying to accuse you of malfeasance, I'm just saying I wouldn't do what you're doing.
Er, what? I didn't say anything about dysfunction and I didn't pat anyone on the back. I did say things about the inherent difficulties of competing in the space though.
Microsoft is known to be quite blog-friendly, much moreso than Google. Blogs like MiniMSFT have been around for years with quite an irreverent take on the company.
Well Chrome does this too, and uses that data, combined with your Gmail and Google Drive info, to provide "better" Google results. I don't really care either way, but let's be a bit level-headed here and accept that the two of the biggest browsers in the market are from companies that run search engines and realize that cross-data sharing is pretty much the industry norm.
"Some versions of Chrome feature Safe Browsing technology that can identify potentially harmful sites and executable file downloads not already known by Google. Information regarding a potentially harmful site or executable file download (including the full URL of the site or executable file download) may be sent to Google to help determine whether the site or download is harmful. Google does not collect any account information or other personally identifying information as part of this contact, but does receive standard log information, including an IP address, URL visited and one or more cookies."
It's a bit confusing for me, it looks like they are collecting visited urls as they are not considered personal identifying information.
I've just learned that with the spellcheck feature they are basically receiving all the text I write on the Internet to their servers.
By the way are we really discussing about the privacy policy of a company that sent its data to NSA? I'm not condemning them, still we should be aware of this and never forget.
Bings' lack of market share compared to Google is not because of any of the reasons listed.
It has to do with the perception people have, and the fact that they more regularly associate searching the web with Google and "googling it" vs. Bing and "binging it".
At least this is the case for average, non-hacker news people.
Why does Google have this perception? I don't know, but I will guess it likely has to do with it being the one search engine that really caught on and was able to sustain its momentum.
Competing against google would not be done by any of these incremental/less important technical metrics but more having to do with changing the notion of searching the web, and how, in a likely very niched way at first.
I am not sure if this is true, but your framing of the problem space is good. There should be much more public discussion(maybe there is) about how to achieve this.
Public perception is also the reason that Microsoft can't hire machine learning and search quality engineers, because the question "Do you want to work at Microsoft on search quality?" starts with "Do you want to work at Microsoft?" and for many people the answer to that question is "no" regardless of the remainder of the sentence.
To say nothing of the engineering challenges with crawling the entire Internet. Your first cut will likely refresh every several months or so. Remember when Google did that, back in like 2005? It was very effing hard then. Google now crawls pages newly linked from authority sites essentially instantaneously.
"Bing exists than it is that it is equal to or better than Google in every way"
How is Bing better than Google (Search) in every way? With worse relevance and smaller index? Although, IIRC, Bing used to train their neural net based on Google search results to match the Google search quality (remember the 'torsoraphy' incident)?
FTA: "social may pose an existential threat to Google’s style of search"
No. Sometimes I want general information: "what is the airspeed of a laden swallow?" Sometimes I want information dependent upon the interests of the hundred or so people I know: "what do you all do in Toledo?" Sometimes I want information dependent upon my interests: "if I enjoyed 'Requiem for a Dream', what other movies should I watch?"
These things conflict: I could not possibly care less what my friends think is the airspeed of a laden swallow.
I use Google for #1, Facebook for #2, and several other sites for #3. Google has (or had) a perfectly viable business providing the best answers to general questions, and it's a shame that they're screwing it up.
Sorry, but you're wrong. Understanding a good amount of detail about your users, and in particular, about their social context, is a massive win for search relevance. Google invested quite a bit in acquiring Aardvark, for example.
So, even if you don't ask your friends about something, the information is still really important.
This. I think people get caught up too much in the small fraction of queries that are facts of the universe ("What's the universal gravitational constant to 3 decimal places?") versus the much much larger fraction that depends upon much larger contexts ("Movies playing in theaters nearby," or "Rock music," etc.)
"I use Google for #1, Facebook for #2, and several other sites for #3."
I think this is the crux of the issue, not that Google thinks social is the end-all answer to everything -- similarly, something like Wolfram Alpha isn't trying to solve the "which restaurant nearby should my friends and I hit tonight?" type of questions. Rather, it's more that Google wants you to be able to approach its products with ANY question or ANY search problem, and it can answer it.
Now the problem is whether or not they can pull that off, or you really need independent separate tools/services for each little question to answer.
Social Search doesn't mean asking what your friends think an answer is.
Instead, it's the most likely approach to questions like "where can I quickly get good pizza?"
A question like that uses social search (beyond your friends!) to find people with similar taste to you, find where they eat pizza and to calculate likely travel time from your current location.
I just checked out Bing with some well known search queries and the results are very close to Google's.
What makes me slightly uncomfortable are
- the name and the branding. It sounds irrational but the name annoys me and I just do not want to type it. Bing doesn't sound sophisticated, smart or anything and the big picture on the landing page is unwanted and makes the product feel unfocussed. Maybe the reason is just that Bing never did something way better than Google and thus, there were no opportunities to load this name with a good reputation with some achievements.
- the speed: Google is (or feels) still way faster here and there (when entering something into the search it instantly switch to the SERP view)
This! The branding of Bing is terrible. When I see the big picture I always think: this has nothing to do with what I'm looking for - and "what I'm looking for" should be the single most important thing for a search engine.
I don't get the point of this post and I think that almost all the lessons the OP wrote down are wrong. Is this post an excuse why MS is performing that bad with Bing? Why is there no one from MS stepping forward and shouting, yes Bing is still behind but we will fight tooth and nail to crush Google -- no, instead some MS employee is writing a non-approved post where he gives 6 lessons why MS sucks and why life is so hard, wtf? Maybe this is the reason why MS falls behind year by year because people behind MS are too laid back and lack fighting spirit.
Nobody knows what MS is today, especially nowadays because MS doesn't have any CEO. We know that MS was the company which controlled the personal computer experience in large parts and we know that this experience was ruled by the web and by search for one decade and there is no excuse that MS totally forgot that search must be a integral part.
101 comments
[ 3.0 ms ] story [ 160 ms ] threadis it mostly natural language processing? matching keyword queries to document corpus
from that query how does one know user's true intent? Is he looking for a wikipedia article? is he looking to hook up? is he looking for porn? is he referring to the gender? is he just bored and trolling?
since google's reach is part of our lives, wouldn't they have a very good idea based on your previous history, what you clicked on, what search results you clicked on, what you typed in google chrome.
couldn't microsoft have pulled off the same thing?
And no MS could not have pulled off the same thing. The data MS has about you is not even close to the data GOogle has on you.
Also, users have adapted and are trained to work with Google nowdays, they memorize how to find particular results on Google (even though Google puts a lot of effort into breaking features like "link:...").
I've never cared about funnyordie, didn't even realise it still existed. Wonder how they picked that, or if other users results differ.
The good thing; when just searched for something about UIView and later type something "populating a table", it'll give me SO articles (and such) about UITableView which is what I meant but forgot to type add ios/or uitable.
The bad thing; when i'm doing research, it'll give me the same stuff over and over again and I have to page quite far and add / remove query info / play with search tools to get to different things which might not be the most popular or might not be what Google thinks I want.
The awful thing; the ads... I use ad blockers, but when I get ads (gmail / some sites), they are always the same. For Facebook I just think their algorithm is broken or maybe they have none (I only get ads about dating, cheating, stocks all of which I have no interest in nor ever shown), but for Google they show things I am interested in but I already know all of them, have been there, knew most of them before those ads and they cannot possibly sell me anything. If I wanted something from them I would go there without those ads, so why are they targeting me like that? Show me ads I don't know and if I can use it, I would click.
I'm technical, but I hear the same annoyance from others; a friend of mine rents out houses; the search results when he wants to research the competition are usually almost only his own sites (presumably because he visits them a lot?) and ads for his own properties. He is very non technical, so explaining this all to him is not really sinking in.
While Microsoft's problems are real, this offers little in the way of solutions. That said; interesting read.
Do you feel something in particular sucks about the article? I'm happy to change it!
That said, I use Bing as a comparison point because it's really the only other point that's comparable in terms of, like, infrastructure and stuff. It's contextualizing more than anything. I might have removed all the references to "Bing", but I don't think it would add clarity.
But I see your point.
In fairness I've tried to be honest about our weaknesses and not shy away from the things that are bad. I think I've been reasonably successful, but of course I'd love to hear from you if you think I'm giving MS too much credit.
What would you prefer? Somebody with no relevant experience talking about the subject? I'm sorry, I really don't understand how actually knowing what he's talking about 'marginalises his credibility'.
As the author wrote, aiming at feature parity with Google is probably a bad idea. DDG seems to do this right: they have some better features and one distinct advantage over Google with their stance on privacy. Perhaps Google will reach a certain creepiness factor with all the collected data they're factoring into search results (e.g. I have this suspicion that content from gmail is used too) and at some point it might work against them even for the average user (or am I just biased?).
One "feature" that Google has failed to do properly is custom "site search", i.e. providing a search interface for arbitrary websites. Perhaps this is another opportunity for a new competitor or even Bing (though their "site search" was shut down in 2011 apparently), there is plenty of demand (and lots of crappy custom solutions).
"When each site understands its own data better than Google, its internal search results will surpass Google’s. Google will no doubt continue to provide better global results, but the two-tiered search would decentralize efforts to improve algorithms."
I see it as one of the Achilles' heel of Google. Even if what they are doing seem impossible to copy.
- [1] Letters from the Future: Challenging Google’s Search Engine: http://blog.databigbang.com/letters-from-the-future-challeng...
This is also where dedicated "site search" providers can beat these search engines easily: they can offer a full crawl and depending on the size of the resulting index, they can bill the website owner (e.g. by storage used in GB or number of URLs).
That being said, I think the only solution to beating Google is developing something disruptive, and I mean that in a very classical and traditional sense (not the way disruptive is used to mean anything these days). It means you have to build a type of a search engine (or search solution) that Google not only has no interest in pursuing (due to conflicts of interest/opposite business incentives), but probably can't pursue, or not without employing dramatic shifts in how they do search (which again, makes it even less likely for Google to pursue and try to beat this new "disruptor").
And no, don't look at Bing to provide that. They're a "direct competitor" to Google, and they will always be, because they want to be "in the exact same business" Google is, and therefore, I don't think they'll ever radically change the way they build a search engine compared to Google.
So what can such a disruptive solution be? I can't say for sure, because it probably needs more factors/benefits on its side than just one, against Google. Social "may" be a way to disrupt Google, although I think Google is already rapidly adapting to this attack vector.
I think another could be "extreme-privacy", which Google wouldn't pursue anytime soon without really rethinking how its search business works.
Disruption also means attacking incumbent companies with much fewer resources through innovative business models and cost structures. Do something that would make Google's billions of investment in infrastructure irrelevant. A P2P search engine like YaCy or Faroo seems to be part of that solution. Perhaps what can/will kill Google is a crowdsourced super-search engine, that ends up surpassing even Google in most capabilities, because Google would be just one company, fighting against the whole resources of the world.
As in classic disruption, you also have to consider this new type of search engine will be much poorer than Google in some competition factors (at least initially). I think that is the "relevance" factor. As the op mentions, it's going to be very hard for anyone or anything to beat Google in relevance. They have too much of a headstart. But perhaps there will be such a thing as "good enough" in search, too. Maybe we're not there yet, but perhaps we will be in 5 years, or 10 years. Until a few years ago we kept wanting faster and faster Intel processors. Now we think mobile ARM processors are more than good enough for most of our daily activities.
A crowdsourced P2P search engine powered by billions of people could perhaps reach that point, too, while offering some of that "extreme privacy" and other benefits that Google can't offer.
Still, how do you get that first "10 percent" of users, that end up changing the paradigm, to start using a service that's much less relevant than Google today? You do it in the same way other disruptive technologies win out against incumbents. You promote the other benefits, that Google can't and won't match - benefits that will be important to those "first 10 percent", until you reach critical mass, and then you should be fast approaching that "good enough" relevance, too, and get even more users after that.
And that's probably because Google is incredibly weak when it comes to patents. This is kind of well-known in IP licensing circles, but it's also evident in their desperation to acquire patents (Motorola, Novell, Nortel... and other patent holding companies that nobody's never heard of).
It may not be in their nature to attack, but even if they wanted to, they are much more vulnerable to a counter-attack.
How exactly are server-side patents enforced? When the hardware on which the infringement is occurring isn't exposed to the public, how can anyone verify its presence? I know that for some (many?) of Google's search patents, there might be an indication on the client side if infringement were occurring, but I imagine that there are also many patents for which this isn't the case.
This has some problems however:
1) As you can imagine discovery may not turn up any actual infringement, making it all a huge, expensive waste of time.
2) The unfortunate reality of patent lawsuits is that if you can't prove infringement just by looking at something, you might as well have already lost. Anything that requires expert witnesses to provide input on often comes down to which sides' witness the jury finds more likable, and that's pretty much a roll of the dice.
I'm no search engine expert but I'm pretty comfortable assuming that Bing would have(and does have)had the same issues with rap genius as Google did. The only real difference is that as the market leader Google had to publicly react to rap genius 'gaming' of their system.
I think what this shows is that yes, this is precisely a product problem because if Bing were at the top we would be having this exact same discussion only with reversed roles.
Correct me if I'm wrong but Google vs Bing is not a lot different than Coke vs Pepsi.
Until we have someone come along with a search engine which is actually a better product, market share will only really be a function of effective marketing.
That said, I think your article does a good job outlining some of the barriers that any potential competitor in this arena should be aware of.
But that's not really so much of a product problem as a problem with the problem itself. In other words, it's not really clear that a better product will fix that, because the incentives will probably always point to doing this rather than not.
No one wants that (except Microsoft).
What we want is for Bing or something else to bring more balance to the search market. Today Google is so dominant that people SEO specifically for it, and Google decides what is and isn't ok to do, and can then decimate site's traffic based on that (as we saw with Rockstar in a recent example).
But if there were several successful search engines with no single dominant winner, things would be better. SEO would matter less because you would need to optimize for multiple targets, so you would get away with less tricks (or you need to work much harder to trick everyone, again with the result of less trickery). And no single company could decide the fate of every other company that needs to get traffic through web searches, which is what we have now.
I want to be "A Pepper" but DDG doesn't always give me results ... Any ideas?
Maybe. Google sells site ads; Microsoft sells Windows and Office licenses; DuckDuckGo sells search ads; Apple sells hardware. Following the money, you realize that Google's goal is to produce search results just good enough to maintain its brand, while driving as much traffic as possible to sites that display its ads. Always ask yourself how your interests differ from your search engine's.
The problem for Microsoft is the same problem that faced Apple. Being "as good as" isn't enough to get people to switch. You have to be 10 times better.
Back when Google launched, it was 10x better than Yahoo. Bing seems to have quickly gotten to 1x, but hasn't moved much further.
Pretty much: Apple was faced with a competitor who owned a vast majority of the PC market. They never took over the majority, but they managed to earn a profit from a minority of people who were willing to pay a premium for computers that worked better than the crapware offered by most Windows OEMs. Windows was cheap, and just good enough to write TPS reports.
Google now offers similar crapware: heavily-gamed and -"personalized" search results slathered with ads and just good enough to be better than most alternatives. There's plenty of room for a company to offer something better: paid subscription search, less-complete search with fewer ads and more accuracy, or something else (I'm not a "10x founder," so I can't think of these things.).
It's not like you need to buy an entirely new computer and learn to use a new operating system to switch between Bing and Google.
We're in the predictable phase where everyone hates google now because they're no longer the next big thing, but I'm not a hipster and I don't think microsoft is now the underdog so I don't get the google animosity and I don't get the sudden warmth towards microsoft.
Quite frankly I fucking hate windows, and I wish it would just go away entirely. To that end I avoid giving microsoft money at every opportunity.
https://news.ycombinator.com/item?id=7011954
https://news.ycombinator.com/item?id=6504324
If the last web site you visited was for the Audubon Society and the next thing you search for is Cardinals you will get a page about birds on the first result, but if you had just recently looked for tickets to a ball game on Ticketmaster you will get the baseball team as your first result.
One of the more fascinating things that DDG's "Dontbubbleme" campaign did, was expose just how big a swing in the results your 'meta data' has. People who are running Chrome and are "logged into Google" get very different results than people who make a query with no context following them around.
In the latter case everyone has to resort to giving you the pages that "most" people clicked on from your geographic region [1].
Bing and Google have pretty much exact equivalence if you remove tracking and geo context. (IP blind / cookie blind / search).
[1] For even more fun, set up proxy servers in various data centers around the country and do searches with no cookies or context, and proxied to different geographies. Very enlightening.
And, what if I could select my settings in a dropdown when searching. So, maybe the people at HN setup a whitelist of 100 of the best sites for programming help. I add this list to my account. Now, anytime I want to search for a programming question, I type in the query, select my custom search type from the dropdown, 'programming questions', and then click search. I'll only get results from those 100 sites.
Google will never implement this for a variety of reasons, so it's room for competition.
For example, reddit has a terrible search engine that users often complain about and comments aren't indexed at all. But searching with google often gives subpar results and they don't take into account things like upvotes at all. (I also think sites like reddit are punished by Google's algorithm pretty harshly.)
This might make the problem smaller and more tractable and a place where you could get an advantage over Google.
The truth behind IBM Watson is that it's based on a methodology that makes it possible to put huge teams working on different parts of the problem without tripping on each other's feet.
https://www.google.com/cse/
how can 2 guys in a garage wipe out AltaVista, Lykos, and whoever else was doing search back then (with lots of smart people, data, resources, state-of-the-art software and user adoption to compete against)?
What is missing is a serious discussion on search quality (which made Google Google).
If Larry and Sergey started today, would they be able to compete against Google by providing better search results, or has search quality reached a "global maximum" of sorts?
Does search still suck?
There's also a pretty big difference in scale. The AV team working on search in 2000 wasn't all that big.
Am I the only one concerned with both the fact that IE tracks its users, and the attitude of this Microsoft employee about it?
EDIT: and let's be realistic. The papers on search that come out of MSR expose way more about our infrastructure than the paltry sum presented in this post.
So there's that.
I don't, btw, thanks for asking.
No it doesn't: https://www.google.com/intl/en-US/chrome/browser/privacy/
Your search history is used (if you have it turned on), but that's just from using google.com for search and is independent of the browser you use.
The means may be different, but the ends are the same: users are tracked all across the Internet.
It's a bit confusing for me, it looks like they are collecting visited urls as they are not considered personal identifying information.
I've just learned that with the spellcheck feature they are basically receiving all the text I write on the Internet to their servers.
By the way are we really discussing about the privacy policy of a company that sent its data to NSA? I'm not condemning them, still we should be aware of this and never forget.
It has to do with the perception people have, and the fact that they more regularly associate searching the web with Google and "googling it" vs. Bing and "binging it".
At least this is the case for average, non-hacker news people.
Why does Google have this perception? I don't know, but I will guess it likely has to do with it being the one search engine that really caught on and was able to sustain its momentum.
Competing against google would not be done by any of these incremental/less important technical metrics but more having to do with changing the notion of searching the web, and how, in a likely very niched way at first.
How is Bing better than Google (Search) in every way? With worse relevance and smaller index? Although, IIRC, Bing used to train their neural net based on Google search results to match the Google search quality (remember the 'torsoraphy' incident)?
> it is more important that Bing exists than OTHER CONDITION
Where OTHER CONDITION is what you've quoted. He isn't calming that OTHER CONDITION is true, merely that it isn't the most important thing.
No. Sometimes I want general information: "what is the airspeed of a laden swallow?" Sometimes I want information dependent upon the interests of the hundred or so people I know: "what do you all do in Toledo?" Sometimes I want information dependent upon my interests: "if I enjoyed 'Requiem for a Dream', what other movies should I watch?"
These things conflict: I could not possibly care less what my friends think is the airspeed of a laden swallow.
I use Google for #1, Facebook for #2, and several other sites for #3. Google has (or had) a perfectly viable business providing the best answers to general questions, and it's a shame that they're screwing it up.
Sorry, but you're wrong. Understanding a good amount of detail about your users, and in particular, about their social context, is a massive win for search relevance. Google invested quite a bit in acquiring Aardvark, for example.
So, even if you don't ask your friends about something, the information is still really important.
I think this is the crux of the issue, not that Google thinks social is the end-all answer to everything -- similarly, something like Wolfram Alpha isn't trying to solve the "which restaurant nearby should my friends and I hit tonight?" type of questions. Rather, it's more that Google wants you to be able to approach its products with ANY question or ANY search problem, and it can answer it.
Now the problem is whether or not they can pull that off, or you really need independent separate tools/services for each little question to answer.
Instead, it's the most likely approach to questions like "where can I quickly get good pizza?"
A question like that uses social search (beyond your friends!) to find people with similar taste to you, find where they eat pizza and to calculate likely travel time from your current location.
What makes me slightly uncomfortable are
- the name and the branding. It sounds irrational but the name annoys me and I just do not want to type it. Bing doesn't sound sophisticated, smart or anything and the big picture on the landing page is unwanted and makes the product feel unfocussed. Maybe the reason is just that Bing never did something way better than Google and thus, there were no opportunities to load this name with a good reputation with some achievements.
- the speed: Google is (or feels) still way faster here and there (when entering something into the search it instantly switch to the SERP view)
Nobody knows what MS is today, especially nowadays because MS doesn't have any CEO. We know that MS was the company which controlled the personal computer experience in large parts and we know that this experience was ruled by the web and by search for one decade and there is no excuse that MS totally forgot that search must be a integral part.