Using cloud architecture and crowd-sourced data on web pages, Wowd, a real-time discovery and recommendation engine would rank pages based on whether users actually visited those pages and would return results from all over the web, not just a handful of indexed pages, reports ReadWriteWeb. Currently it is in its private beta version. The question is whether such an approach succeed?
Before I proceed to comment, here is a video interview with the CEO of Wowd conducted by Tim Reha, in case you are interested to view:
In my opinion, the biggest hurdle in the success of Wowd would be the fact that it is a downloaded application and you have to download it and install it on your computer, though thereafter it will run in a browser. Moreover, your browsing history is saved to recommend more personally relevant and interesting content. The moment such a user, who has installed Wowd, visits a web page, that page gets a positive vote on Wowd. Therefore, the number of users (who have installed Wowd) visiting a web page would decide the rating or ranking of that web page in the eyes of Wowd. The keywords, PageRank, back-links, etc., would not be directly relevant.
The first question is how many people will actually download and install the Wowd application? If the user base of Wowd does not pick up substantially and remains comparatively small, then its ranking of the pages would not reflect the real position and perhaps could even be manipulated due to small user base. What is a small user base has to be seen in the light of the vast number of people who visit the Internet everyday.
Then there are privacy issues. Even though Wowd claims that the personal information is not stored on a centralized server, yet the fact remains that the browsing history is saved to recommend more personally relevant and interesting content. How many people will trust a new company for their browsing data and moreover, what is the return for sharing such data? When relevant search results can already be viewed by just visiting an established search engine such as Google, how many people will bother for installing an application and share their private browsing data to get skewed search results?
The next important issue is the fact that old web pages will get higher rankings while the new pages will start with zero rankings even if such new pages are more relevant and provide better contents. By the time a new web page on a particular subject or keyword has been added to the web, the old web pages on the same subject or keyword would have already assimilated positive votes on Wowd from the users who have already visited those pages. It means that the new pages would always rank much below the old pages in the search results irrespective of the quality and timeliness of their contents. It is like a chicken and egg story. Older pages will get more views and hence higher rankings while new pages will have to wait for quite some time until the publisher tries some other tricks of pushing his pages. But, the organic method of getting to a new page through search results would not work efficiently. Compare it with the present situation where Google ranks new pages almost instantaneously the moment they are published and many times even a new page can begin with good ranking in search results depending upon its contents and other relevant factors. Moreover, now Google provides options to view the search results from different time horizons, say last 24 hours and there are tricks to view the latest pages added to the web down to last one second!!!
Another issue that I am not able to comprehend is – what will actually happen to a new page? Will it not be listed in the search results till it is viewed by at least one user who has installed the Wowd application? Does that mean that Wowd will not be crawling (like Digg) the web in search of new pages? Or alternatively, will a new page be listed in search results with zero rank? It appears to be a grey area and perhaps a weak link in the logic of Wowd. I say so because Wowd claims to be a real-time search but its logic with regard to new pages appears to be flawed. Is it not self-contradiction?
Yet another problem with such flawed search engine would be that while initially some people may use it to search out of curiosity; but, later, when they don’t see the quality and quantity of the search results, they are likely to forget it in little time resulting into possible unnatural end thereto due to the vicious cycle effect.
While one has to wait for some time till their private beta version produces its results to see how relevant the search results of Wowd are vis-à-vis the established search engines like Google, I’ll not hesitate to stick my neck out and put it bluntly that I am not very optimistic about Wowd being in a position to give any meaningful real competition to the established search engines in its claims of a real-time search. Anyway, meanwhile, Wowd and its promoters / founders can try to derive the maximum benefit of the venture capital that they appear to have secured already. Sorry, Wowd, for my frank comments. I’ll be really happy if you can prove me wrong by your performance as and when your search engine is launched to the public in future.
Did you like this article? To get all such articles from Tech Superb directly:
Related Articles:
If you would like to make a comment, please fill out the form below.
Great analysis, your thinking is great for most points, but Wowd is different from what you think of it, in many points it is actually on your side of the argument.
Here is a quick wowd demo, that will explain many things:
http://www.youtube.com/watch?v=usvL2qoe75Y
I will also explain few things:
- You can use wowd hotlist and search from the website, but for all other things you must download it.
- Popularity is just one (important) piece in the formula.
- You can choose if you want popular or if you want fresh results. Time from the point when you nominate page for including into index, until anyone can search for it, is measured in seconds or minutes.
- Your webhistory is stored locally on your machine, and recommendation is using local statistics from that history to find good pages for you. So all personal information are stored on your computer and nobody else can access them.
You can leave an email on the wowd.com, and get an invitation for private beta, to see for yourself.
How can we get invitation key?
Igor, thanks for dropping in and explaining the details of Wowd project. I extend my best wishes for the success of the Wowd project. Let the users decide how successful it is going to be. Its success will definitely be useful for the people for the simple reason that competition is always good for the users and a good competitor can lead to improvements in the services of a rival company too.
Jasim, Wowd is presently in the private beta version so it is open only to the invited persons. However, to get an invitation key, you may visit Wowd site at http://www.wowd.com/ . You may enter your email address on that site to receive the Invitation Key to join and try the private beta.
My name is Borislav Agapiev, I am the founder and CTO of Wowd. First, let me thank you for your feedback, we find it invaluable when people express their opinions on the topic of search, we are very passionate about it.
I wanted to clarify and respond to some of the points you mentioned. The first question you raise is how many people will download the app. It is definitely a valid question and the ultimate outcome remains to be seen.
The number of users on the Internet nowadays is huge (well over 1 billion), and there have been many examples of of passionate and motivated groups and user segments that were (and are) willing to try new approaches, including downloads: SETI@Home, Folding@Home, Grub (distributed crawling), Majestic-12 (centralized search with distributed crawling) etc. In addition to these distributed computation approaches there are also many more segments where people feel passionately about causes such as privacy and concetration of too much power in various corporate entities, green issues of using computation resources efficiently, also people who want o be part of significant game-changing movements such as social search etc. The point is that there are many groups of users who have already demonstrated keen interest in causes related to what we are doing. We definitely want to take our message to them.
This is not to say that we do not envision Wowd as being very useful for mass consumer audiences, we definitely think that our approach will appeal to them too but indeed we will see how many of them will be willing to do a download (initially). It is also true that early adopters and power users often lead the way for others.
Another important point is that because of our distributed cloud architecture and efficiency of our implicit voting and ranking system the number of users needed to reach critical mass of producing something really interesting is not that high. Consider, as an example, a collection of 10K active users – the aggregate RAM in the system would be 2TB and disk resources would be in 100s of TB. As a point of reference, it is pretty well known that a world-class general search index is on the order of several hundred TB. I can say that provisioning such resources for a search startup is not cheap nor easy and very few are even attempting it. Now consider such a system with 100K, or 1 million or 10 million users or more, our distributed cloud architecture can scale to any number of users who decide to join. Of course, the computation and attention frontier power at such scales would be truly staggering.
The next point is how efficient in terms of voting Wowd users would be. Consider Twitter as an example – they get about 2K urls/min (before deduplication or any quality/spam/relevance analysis) from their user base of 40M+ unique visitors/month. What this means is that very few Twitter users bother to post links – yet the results are already pretty amazing, in terms of how quickly interesting new things show up in their results.
A system which is optimized for using user actions as ranking signals, such as ours, will be able to leverage implicit user actions (ie clicks on links) (much) more efficiently since every user generates many more such events naturally, during browsing, with no explicit action required.
There is one thing I want to clarify, you seem to imply that with a small user base, there would be no ranking information available to rank results. That is not the case, with a small (or even no) user base, it is possible to rank results using well known ranking algorithms and signals such as link analysis, keywords in titles, anchors, titles, keyword densities etc. So our baseline is not that we cannot do any ranking, the lowest case is actually conventionally available techniques and ANY user data will only IMPROVE it.
You also mention that we will not be able to rank new pages and it is not clear how to rank them initially. First, that is an issue for all search engines, including Google and there are many ways of handling it. A simple answer is that there are other ranking signals that indicate quality and relevance of new pages, there are also other indirect methods relying on reputations such as authors etc. So it is quite possible to assign meaningful initial ranking to new pages. Another question you raise is that it would take too long in general for popularity counts to propagate – that is not the case, as can be seen by an analysis of Twitter and other attention based data streams. In a nutshell, when new things break out, things tend to happen fast, that is indeed the point of real-time search.
I will stop here, this discussion has been quite interesting and I will post a related entry on my blog http://distributedsearch.blogspot.com, thank you for starting it
Borislav, thanks for clarifying certain issues relating to Wowd. From what you said, it appears that your search system will be using the combination of user-ratings as well as the conventional page-ranking algorithms. The new web pages will be ranked initially using the conventional algorithms for ranking pages. It should take care of the concerns relating to ranking of new pages. The real difference will of course be the scale of number of users who rate the pages. If the number of users rating the pages using Wowd is really huge, it will definitely add quality to your search results; in the opposite case, your search system will be like other conventional search engines. My best wishes for your search system.
Secondly, though it is at a different place altogether, I may also mention that Google has just launched Sidewiki which will allow users to offer comments on any web page which will appear on the side bar in the supported browsers through Sidewiki-enabled toolbar (see, my article “Google takes control of comments on all websites with Google Sidewiki”). In one sense, it will be similar to your project in that it will allow users to comment upon (thereby, indirectly rate them though there is no direct rating system by users as in your system) any web page on the Internet. However, the similarity ends there. Because, you will be using the user-ratings for indexing the search results, i.e., “before” the search results are produced and how they are ranked, while Google will be showing the user comments on a web page “after” a web page has already been “opened”. But, yet Google Sidewiki will also be in the nature of user-voting on a web page (though indirect and at a different stage). One has to see how Google Sidewiki performs. The obvious advantage to Google is its large user base for Google Toolbar and moreover, while new downloads of toolbar will be Sidewiki-enabled, it is understood that Google will be automatically upgrading even the existing Toolbars over a period of time. So, this huge user base will make a difference and so will Google’s brand value. As I mentioned, it is not directly relevant for your project but it is going to at least indicate how much interest users can show in rating a page or commenting upon a page if a project can have mass appeal to users.
Borislav, thanks again for visiting and wish you all the success in your project.