Google Shopping aka Google Product Search aka Froogle

Google’s commerce marketplace is under attack by Amazon and Facebook causing Google to respond. Google+ and its current experimentation with Product Search – its shopping aggregator vertical – are visible reactions.

Changing the business model and experience of a shopping aggregator site is a dramatic event. I have driven MSN Shopping and Bing’s commerce vertical through several such cycles from paid placement, to free, to pay-per-click, to pay-per-action and from 5 million paid listings to 100 million free and back down to 50 million listings.

The first step is to have deeper relationship with merchants that want to engage with a new listing product. Google has announced that it will transition its catalog from free to paid inclusion. This is a big change since it will reduce the number of merchants in Google’s catalog significantly; kicking out niche product providers and feed aggregators that constitute the backbone of any product catalog’s tail selection.

The next step is to establish a merchant portal that allows managing product data and merchandizing to shopping customers. Google has recently acquired Channel Intelligence a commerce enabler that offers feed generation, optimization, reporting and consulting.

The last step is to introduce a shopping experience that allows merchants to compete for customers by actively controlling how and where their product appears.
Ranking in Google Shopping, when the full transition is complete this fall, will be based on a combination of relevance and bid price–just like Product Listing Ads today. This will give merchants greater control over where their products appear on Google Shopping. Over time they will also have the opportunity to market special offers such as “30% off all refracting telescopes.” Google Commerce

Google’s move will alienate small merchants and has the potential to reduce the satisfaction of its customers through reducing selection. There is nothing that undermines confidence of a shopper in an aggregator more than not finding a product in the catalog or finding it on a competitor’s service.

On the other side Google’s move has the ability to greatly enhance the shopping experience and thus to re-vitalize the shopping comparison market, which has become a little stale especially with discount and flash sites stealing the show.

Two-sided Marketplaces

Consumer online services are driven by commerce, i.e. they are intermediaries that bring buyers and sellers together.

Portals (Yahoo, AOL, MSN), Search (Google, Bing), Social (Facebook, Twitter, Pintrest), Local (Yelp, Google, Bing) and Commerce (Amazon, eBay) are all platforms that create value by enabling buyers and sellers (advertisers) to interact.

Two-sided marketplaces are fascinating because of their network effects either inside each participating group or across groups.

They are the reason that the freemium model exists. Marketplaces allow subsidizing one side – typically the buyers – by charging the other side  – the sellers or advertisers. Network effects also create a winner-takes-all effect that allows one platform to dominate.

Google with its auction-driven advertising platform and its Internet properties has become the largest consumer marketplace. However not all is well in the world of searches with “purchase intent” and Google’s monetization of the last click.

When I was running MSN Marketplaces and Bing Shopping consumers would use comparison search engines to look for products and compare prices. If Amazon was close enough with its price they would head over to Amazon and buy it there – a simple matter of trust and convenience associated with the brand.

Price are much more transparent in todays online market and competition has forced retailers to price match in near real-time. Today customers might as well go directly to Amazon’s site and search and purchase the product there – bypassing Google as intermediary.

The net effect of this is that more buyers are heading directly to Amazon today. This brings positive network effects into play that pull more advertisers into Amazon’s marketplace and ad exchange, which in turn attracts more buyers and so on.

Amazon “database of individual purchase history” beats Google’s “database of search purchase intent” hands-down in its targeting capabilities and therefore attractiveness to advertisers.

Amazon however is not the only one eying Google’s market position.  Facebook’s has created successful advertiser marketplace driven by its massive “database of connections” with its rich profile and interest data. Facebook’s understanding of users and their activities eclipse Google’s knowledge derived form anonymous searchers and their click.

Two-sided marketplaces can create phenomenal growth if their network-effects are positive. The flip side is that negative network-effects are equal strong causing buyers or advertisers to withdraw stripping always Google’s economy of scale and pricing power.

This competition will be interesting to watch and study.

Like Economy

The Open Graph protocol allows Facebook to aggregate social activity across the web. The protocol and the like button are highly successful examples of a centralized architecture that generates economic value through decentralizing social actions.

lice economy

The Open Graph protocol defines mark-up (meta tags) that turns web pages into Open Graph objects that are understood by Facebook’s platform. A publisher participates in the like economy by annotating his web pages with meta-tags of the protocol and by hosting the ‘Like’ button plugin on each of these pages.

A user who “likes” a page creates a connection between “his” node in the social graph and the open graph object extracted from the page – example objects are movies, sports teams, celebrities, and restaurants. In addition, the action of ‘liking’ generates an entry in the user’s news feed, ticker and timeline, along with a post of the web page’s link and image.

Facebook’s model of value creation is social amplification and hinges on information dissemination between networked users, i.e. users and their friend in the social graph. The value of a like is its ability to reach the social feeds of friends and trigger further attention and activity there.

The value calculation algorithm of the like economy is Facebook’s EdgeRank, which ultimately decides what is to be shown in a users’ social streams and when it is to be shown.

Facebook has created the Open Graph ecosystem from scratch – a significant achievement. The system is still evolving and I have therefore focus this discussion on the like connection in the open graph.

The positioning of the like button as an one click expression of affection for a real-world thing and its ability to generate social capital by bringing the sentiment to the attention of friends is innovative. It reduces the barrier to participation for users significantly compared to a rating system (e.g. star or positive/negative) that forces the user to make a value judgment.

The problem with the current implementation of the like button is that it creates a permanent connection in the open graph. Affection changes over time and as user I would expect that there is a decay associated with a like that makes it disappear over time.

Any kind of search or discovery algorithm that uses likes as signal will have to factor this in. This will make it confusing to users. It would have been better to split the functionality into a like that is transient and functionality that allows users to save what they have liked.

The innovation on the publisher side is the connection of user profiles to brands based on social activities rather than page impressions created by anonymous clicks. The ability to target networked users based on rich demographic data extracted from profiles has the ability to change commerce.

Facebook’s recent Open Graph Search announcement will make the like economy complete by offering a service that makes the massive aggregation of connection in its databases useful for users and advertisers the like.

Before the announcement the like button’s main value proposition was tracking and analytics. The impression logs it creates are of significant value but privacy discussions and public attention to data collection has made it hard for Facebook to capitalize on this opportunity.

Open Graph Search has been discussed widely in the press with a broad range of speculations about its focus. It is interesting to look at three opportunities beyond a Facebook site-specific search:

Web Search Service such as Google/Bing

20% of the world’s webpages have the Facebook like button installed according to w3techs. This a huge achievement but a far cry from the number of sites Google or Bing indexes. Search engine exploit to hyperlinks that are intrinsic to the web, while the like button is a Facebook specific extension. It is therefore hard to see how the like signal could scale to compete with Google/Bing in relevance and especially in relevant content in the tail. You can find a discussion of the link economy and search here.

80 of comScore’s U.S. Top 100 websites and over half of comScore’s Global Top 100 websites have integrated with Facebook in 2011. The top sites are to large degree commerce focused.

The aggregated like information is a viable signal for online commerce, which is trend and popularity driven. Facebook could therefore successfully enter the market with an innovative marketplaces engine driven by like activities of its networked users on participating commerce sites.

Local Directory Service such as Yelp

Facebook’s Open Graph aggregates many objects such as restaurants and services that are typically found in local business directories. In fact Yelp, the most prominent service participates in the Open Graph, i.e. annotates its web pages with meta-tags making its information available in Facebook

However good ratings systems provide information about the characteristics of products and services rated. A simple like connection or a count of likes would be a meaningless flat rating, preventing users from making any kind of decision other than that of a trend.

Designing a good reputation system for ratings and reviews is challenging and most systems are vulnerable to gaming. The social graph could be a good starting point to innovate in reputation systems.

In my time running MSN Shopping we have invested heavily in opinion extraction technology to factor reviews into features and opinions and to aggregate them – the technology exists. However it hard to envision that likes in ones friend circle could aggregate into a comprehensive set of local business rated by trusted experts that are also friends.

Discovery Service such as Pintrest

The like button as fleeting sign of affection for a thing is a great discovery signal. It is designed to draw the attention of friends to something a user has discovered and as such could be used as a signal to create a relevant browse experience for web things.

A discovery engine for web things would work similar to a collaborative filtering system as used by Netflix or Amazon, but would use the open graph to compute recommendations for a networked user.

Pintrest and its many clones exemplify good discovery experiences that could be driven off the Open Graph and it addition would offer a place to integrate the sophisticated refinements Facebook has shown in its search previews.

As stated above a large percentage of commerce sites integrate with the Open Graph protocol opening the door for well-known monetization models for a discovery engine.

Google versus Bing

“Google and Bing are the two most popular and widely recognized search brands in the world. … Although when compared in double blind search tests, the majority of people could not tell the difference in results, people will swear they love one or the other.” wikivs

As a simple exercise to make my point I have replaced Coca-Cola and Pepsi in above wiki entry by Google and Bing respectively.

Today there is no benefit anymore of using the two major search engines in parallel to validate results or to check that nothing has been missed.

It is impressive that Bing did catch up to Google with its first-mover advantage and its economy of scale. On the other side it is equally impressive that Google has maintained it market position considering ongoing technology shifts such as mobile and social – no second-mover advantage for Bing.

Google’s dominance in the market has introduced a state in which the user experience and result sets of search engines have become very similar. Any difference in size and freshness of crawled index of web pages are so small that they are hard to experience during normal usage.

I am a heavy user of search and I would like to see some innovation in the way search could make it easier to find content in specific domains introducing serendipity at the same time.

Currently query understanding and result ranking are one-size-fits-all independent of any understanding of the domain you are searching in. If you are a coffee lover and want to search for Java you will get drowned in links for Java the programming language. If you are a technician and are looking for information about seals you get presented with the singer and cute animal pictures.

In both scenarios the user has to understand the language of the domain he is searching in to disambiguate the query and find what he is looking for, i.e. Java coffee or engine seal. This is hard especially if it is a domain the user is not familiar with.

By providing domain specific language models search could make it much easier to find information and to explore a new domain. It also would allow to find links in the tail of a domain that otherwise would be drowned in general noise.

In my time at Bing we did developed sophisticated language models for commerce search to help shoppers to find what they are looking for. The investment in technology was significant and obviously it makes sense for high value commerce queries.

Search engines in order to differentiate could optimize for financial, government, engineering, religion etc. domains and enhance search by making query formulation, disambiguation and result filtering easier serving content that currently is inaccessible.

Three Screens

I have been always fascinated by the idea of connected smart devices, i.e. TV, tablet (as replacement for a PC at home), and phone integrated by a common operating system that makes content interactive, i.e. blurs the line between content and apps.

Such a connected smart device platform covering the three screens would have many attributes of Apples iOS on iPhone, iPad, and iPod such as the hardware and app ecosystem. TV however is a different device – it is multi-user – and would need to be integrated in an intelligent way into such a system.

Making interactive content available – think multi-player games on a game console – would open up a complete new dimension for commercial and information apps.  “Gamification” as strategy to enhance user engagement would become truly meaningful.

In a three-screen world an app developer ecosystem such as Apple’s would unlock the developer creativity necessary to produce the innovative content necessary to convince advertisers to invest and consumers to engage.

Reality however is that the TV has already lost being the “first screen” in the attention economy. People are wedded to their smartphones even at home and the device has become the entry point to social media, e-mail and apps (e.g. games).

Sooner or later an attention deficit to TV will change the behavior of advertisers and associated money flows. What comes next for today’s TV content has been illustrated painfully by the newspaper and book industry.

Coming back to why I would like to have access to the TV as app developer – commerce  - reaching people in their living room with online shopping services.

Shopping on the web has become boring.  Online an invisible crowd is going through a similar shopping experience in parallel. Through their collective activities analyzed by sophisticated data mining algorithms they are influencing our shopping experience and making it predictable.

Portals, search, comparison-shopping have turned advertising into links plus text or even worse annoying animated gifs. Targeting ensures that the same boring advertisement can follow us everywhere on the web and into our social feeds.

In the physical world window-shopping with friends is a fun pastime and presenting goods in an appealing and clever way is a good retail strategy. Translating such an experience online and into the living room requires the TV.

There is enough cool retail/advertising content out there but online it has been overtaken by links and harvesting of commercial value by convincing online users to click on them. We have developed a platform to change shopping on social and mobile as part of davai, but tablets and phones are still mostly solitary devices.

Commerce apps on TV would allow telling an engaging stories in TV style, with direct user engagement on social and click to purchase as in Web commerce.

Link Economy

“Search engines like Google interpret links to a web page as objective, peer-endorsed and machine-readable signs of value. Links have become the currency of the Web. With this economic value they also have power, affecting accessibility and knowledge on the Web.” Links and Power: The Political Economy of Linking on the Web

The interpretation of content on the web as a link that can be exchanged for value is a powerful abstraction and it helps understanding search, social media and the ever-evolving shape of the web. It is also the ultimate manifestation of how the web has commoditized content and it’s publishing.

Publishers create links by publishing content on the web that can be hyperlinked. Links can be followed, aggregated, shared as well as transformed and their value can be calculated. Links pass value between publishers.

The engine to calculate the value of a link on the information web is Google. Google by attaching its engine to hyperlinks – the fabric of the web – has become the dominant link aggregator.

link economy search

The external representation of the value of a link as calculated by Google is its position in search results. Today’s search engines use very sophisticated machine learning algorithms that are trained to determine what’s valuable and what is not. The original Page Rank used by Google has become one signal among numerous.

Through its auction model Google allows publishers to bid for a more prominent positioning in its result sets, i.e. publishers can increase the value of their link on the web by paying Google.

As a note aside: link aggregation has centralized the de-centralized organized web in the index of search engines; think about the impact of search engines being down.

The commercial impact of link aggregation has spawned a whole industry centered on SEO/SEM and their different tactics to increase the value of a link. The commercial impact also has started to cause problems in the ecosystem in form of link farms and exchanges that try to game search engines.

The market of links created by search engine is efficient today. SEO and SEM are well known techniques and service providers are readily available and tools such as Google analytics have created transparency. This means that it becomes harder for all participants to find additional value beyond what is extracted today.

  • Search engine will have to find new areas of growth. Google for example has announced the transition of Google Shopping to a paid inclusions model – publishers pay to have their link in Google’s result set.  This experiment is a profound change and has to be understood as a step to capture more of the value of commercial links with obvious implication for other verticals and web search.
  • SEO/SEM is a must for online retailers and it has become hard to turn it into a competitive advantage today. One reason I believe that online retailers have to look elsewhere for low-cost traffic, i.e. find ways to better engage on social.
  • Traffic arbitration, the buying of low-cost traffic (keywords) on search engines and selling it at higher price to other sites, will see diminishing returns, which will hit comparison-shopping sites.

The web is rapidly changing from being just a source of information consumption to a place where people produce, consume, share and interact. The social web started changes the link economy as defined by Google and it started with Blogs.

Blogs have always been a unique part of the link economy besides offering users the ability to become publishers. Blogs created a loose system of value exchanges through reciprocal linking.

It is an interesting point of view [pdf] that Google decision to no-follow links in the comment section of blogs has been the starting point for a complete new set of link discovery services such as bookmarking and sharing sites.

Platforms such as Facebook, Twitter and Pintrest have taken link sharing to a new level with their social buttons and user specific activity feeds. These new players are becoming link aggregators themselves thus are changing the way links are valued.

A second force is changing the link economy – mobile. Mobile with is native apps as content representation breaks the hyperlink model and in doing so disrupts the link economy.

Apple by Design

Apple Mac’s and i* devices have the tendencies to pull their sibling products into our house. I am using an iMac with a second Thunderbolt display to do my work. I can produce content as well as work at interface design and it does the server part too, although I still have to have Linux servers for scale reasons.

Using an iMac it was an obvious choice to get a MacPro to continue working while traveling.

Our infrastructure had been linked together by Netgear wireless routers but when the iPhone came along it would not release its IP address while sleeping hence the router would assign it twice and crash. Out went the Netgear and in came the Airport Extreme with is easy configuration utility.

The next question was what tablet computer to use. It wasn’t really a question since I mainly use it for reading and our offspring watches movies and plays games on it. iTunes and AppStore makes it possible to share content across all devices – cool.

Getting the Apple TV basically was a no-brainer at this point although I am not super impressed by its current incantation. The last step came when I needed a better backup solution and again the obvious choice became Apples Time Capsule with its integrated Time Machine experience and gone was the backup server.

Where I am going with this is that Apple gets it – devices are not only beautifully designed and come packaged to make you feel good – they integrate well right out of the box.

I can control the home network using the AirPort Utility on PC/Laptop/tablet, I can share my Browser reading list across all devices using iCloud, all movies and music tracks are available on all devices including the audio system, and I can switch from PC to tablet to phone and can access most of my stuff.

Not all is perfect in the Apple sphere through and more complicated scenarios require fiddling.  Connect your iPad or iPhone to iTunes and a backup / restore / sync / ”whatever process” starts that runs forever – I still haven’t figured out how to change it. It is my fault for not learning the environment but why has it to be so complex and take so long.

Try to find music and movies in iTunes – I still get lost in different libraries and have to explore drop down menus. And why is there a different AppStore – of which I like the experience versus iTunes.  Apps, movies, music are content and Netflix has shown how a content discovery experience should looks like.

If Apple cracks the TV by merging programmed content, on-demand content and games with an ecosystem for app developers I could see some very interesting application scenarios. Google, Microsoft or Samsung are candidates for such an innovation too but I doubt that they can create an app developer ecosystem like Apple.

Mutitouch Gestures

“I’m not against touchscreens as a rule, that is, except when they’re mounted vertically in front of me, ready to get covered in fingerprints, make my arms ache, or wobble about when I prod them.” DigitalTrends

I don’t want to touch the screen of my PC or laptop either but I am very into multi-touch gestures controlling my devices and computers. The iPad has brought it to the forefront in my mind how efficient it is and how intuitive it can be.

I started to wean myself of the mouse during last year and I am now mouse free. This has been a painful process at first but now comes natural especially crossing all the different devices.

Find the different tap, swipe, pinch and scroll gestures on a Mac here. I haven’t used a Windows 8 touchscreen version “in anger” but I assume it functions very similar.

Of the three screens I only want to touch the tablet/phone – PC, laptop and TV are off limits. For the PC and laptop I want multi-touch control using a touch pad, which brings me to the question of how one wants to control the TV.

Kinetic I assume could be an alternative to TV’s with touch control, but I have a hard time to see folks gesturing in the air in order to change channels or adjust sound.

My preferred solution would be to use my tablet to control the TV, one reason I am waiting for the much-ballyhooed Apple TV, to see if they can get the human user interface right.

On a note aside it is actually very difficult to integrate touch (as of touching the screen) into a programming world that has been mouse driven – the distinction between hovering over something (mouseover event) compared to actually clicking on something (mousedown plus mouseup event) does not exist on tablets/phone.

I.e. having a pointer is a key part of the PC user experience design and does not translate onto touch devices. This was one reason for the spat between Apple and Adobe about the migration of Flash content to Apple devices – the content simply would have not worked and performed and this would have reflected badly on Apple’s devices.

Adobe Flash apps that are developed for mobile using multi-touch are fine as the ongoing development of Flash mobile games have proven. However the lack of migration from web to Apple’s devices has undermined Adobe’s developer ecosystem. It will be interesting to watch how this transition will pan out for Windows’ developer ecosystem.

Online Ecosystems

Every online service has to think about how to attract users and to create an ecosystem. The best strategy to grow a user base is to attach to an existing ecosystem, i.e. build your service on top of a popular hardware, software or content platform. Once a new service has been established it can expand its own ecosystem to maintain growth.

There is a broad spectrum of ecosystems online: search (Google and Bing), social (Facebook and Twitter), mobile (iOS or Android), games (Apple Game Center or some hardware consoles) to name only a few.

Search

SEO/SEM companies that provide services to optimize the value of algorithmic and paid link dominate the ecosystems of search engines. This includes data feed aggregators that target verticals such as shopping or local.

Search engines have become the access gate to content on the web but none of the players has developed a vibrant developer ecosystem for value-added apps. Surprisingly this strategy has moved unchallenged into the mobile world.

Both Bing and Google have efforts in place to make APIs available to startups to develop innovative solutions on their respective platforms using data amassed in their indexes. However without a strategy of how value is created by third-party apps it is hard to see how the current situation changes.

Currently the only bet to leverage search is to publish links to web pages. Content engages users to join the service or in case of mobile to install the app. In my time at Bing we did develop an innovative ecosystem of 3rd party search apps that was never productized. Opening the search experience to apps developed by 3rd-parties is challenging and I would hold my breath that we see it happening.

Social

As part of davai.com we developed an innovative way to generate merchandizing apps that created living product stories to reach customers on Facebook and Twitter.

Facebook is an attractive ecosystem to attach to because of its large number of users. The Facebook APIs especially Facebook connect are stable and documented well but complex and require significant investment as well as a good API library.

Personalized apps and games that are developed with networked users in mind are benefit the most from Facebook’s ecosystem. However there are no low hanging fruit anymore with 9 million apps competing for attention and Facebook cracking down on apps that it believes exploit its ecosystem.

While Facebook has opened its walled garden Twitter has gone the other direction and pretty much locked down its ecosystem. At the beginning of Twitter its ecosystem has been phenomenal. Innovative clients and platform extensions popped up everywhere to fill in gaps that weren’t addressed by Twitter and its focus on simplicity.

Twitter initially looked like a great place to build a social commerce engine that would automate encounters between consumers and advertises. In our approach we used structured activities to overcome the noisiness of the twitter stream and allow users to self-identify their brand or product affinity. Scoring and clustering using the implicit and explicit social graph would identify the audience that could be targeted by advertisers.

The challenge to keep heterogeneous clients in lockstep with an evolving platform and the push to monetize its service make Twitter’s strategy shift understandable. Twitter has introduced structure into its stream which make its service much more useful similar to our approach, but by making its developer ecosystem less attractive it is becoming a centralized content broadcast service mainly betting on its own ability to innovate.

Mobile

On mobile, iOS is still the ecosystem to bet on, because of its popularity, i.e. “it came to iOS first,” or “it’s iOS-only”. The ecosystem is unique with its explosive growth, its combination of innovative design and devices features and the fact that people are willing and have been trained to pay for apps.

The challenge of Apple’s ecosystem is app discovery, especially for apps that charge more than $0.99 making a successful launch hard to predict. Apple also has a very strict and time-consuming approval process for its app store adding to the unpredictability of the process for an app developer.

Android is open and developer friendly and, if a cross-platform development environment is used, the more efficient platform to develop against. The “openness” however, i.e. the ability of hardware manufacturers to customize the system, has led to the challenge of Android’s being the most fragmented application ecosystem.

Today being on mobile is a must for any innovative online service. This makes it necessary to deal with the app discovery problem inherent to mobile ecosystems.

Integration with social – i.e. Facebook – can address the discovery problem by employing networked users to pull in their friends by means of app related social actions. Apple’s Game Center, which has become a successful discovery mechanism for games on iOS with its friend lists, leaderboards and achievement mechanism demonstrates this approach successfully.

The remaining challenge is the integration of mobile apps with the search ecosystem especially if a service is developed with a mobile first strategy …

Structured Data Search

Searching structured data is hard. At first glance it is reasonable to assume that the additional information provided by structure – called meta-data – would make it easier to provide relevant results, but is not.

It speaks to the strength of Page’s and Brian’s approach of ranking based on link authority and its current evolution that web search relevance is still hard to beat.

By 2020, there will be 4 billion people online creating 50 trillion gigabytes of data according to IDC and most of it is structured (or semi-structured to be precise). I continue believe that the next technology wave in search will come from addressing the structured data search problem.

One useful classification of structured data is the way it is generated:

  • Community generated – Wikipedia as the crowd-sourced encyclopedia is the number one source of topic specific structured data with others examples being IMDb a movie database,
  • Expert Curated – examples are Wolfram Alpha’ dataset or more domain specific government databases such as NASA’s NSSDC, and
  • Proprietary – for example commercial product and services catalogs.

One challenge in using and aggregating structured data is authority of the source. Google for example doesn’t list in its topic summaries the source while Bing does. In both cases the entry is most likely generated from a mixture of homegrown and external data sources.

The more structured data is pulled into the experience the more relevant the question of authority will become. As a user I simply would like to know where the underlying data – for something that is presented as a fact – comes from.

The second classification of structured data is based on how it is used in search:

  • Understanding – extracting entities and relationships with known semantics from a dataset is the base for query-understanding and disambiguation, and
  • Aggregation – summarization of data from one or more sources to serve it in the result set, and
  • Navigation – of the graph from one entity to another following semantic relationships.

I like to think of structured data less as an enhancement of the SERP – a very domain specific approach once you know what a user wants (i.e. a vertical) – but rather as help to find additional interesting information (in the tail) that is more comprehensive in dealing with what I am looking for.

What I would like to see is the uses of the entities and relationships of the knowledge graph to identify and rank authoritative sources about a topic rather than to pull topic-related information from a few sources into SERP, sources of which I can’t control the authority.

Clicking on a link is easy; determining what is relevant and authoritative for a query is hard and time consuming – that’s where algorithms could shine.