Lumi, a discovery search engine

Web search is changing. Since the start it has evolved both from purely technical point of view (the continuous update of the algorithm and the very appearance of Google) and from a broader point of view – for example, personalized search has raised questions about the neutrality of the information conveyed – hence the filter bubble debate).

Searching on web is getting more and more precise. The algorithm is continuously being tweaked and getting more and more intelligent. The monopoly of a search engine like Google also makes people a bit uneasy as there is too much personal information in their control. Read more here to find out how other search engines are working around this problem.

The growing power of Google has also led to rather firm reactions from competitors – Bing and Yahoo! all – and to the birth of interesting “alternative” search engines, whose strength lies mainly in the respect for privacy (one example: DuckDuckGo). There are also engines devoted to all the queries you’ve done previously, as SeenBefore. In short, the landscape is extremely varied and constantly changing.

In recent times, in particular, we have seen the emergence of non-linear ways of exploration and discovery based more on spontaneous proposals rather than on the mechanical response to a series of keywords.

The most significant example is certainly StumbleUpon, an add-on for browsers that goes hunting for web pages deemed relevant to the user, according to some areas of interest set forth above, and exploiting the likes carried out on various proposals. (Another interesting case is

Here, Lumi works in the same vein.

The project is managed by the two founders of, and re-used its ground concept: the famous service of music recommendation is based on scrobbling, or storing songs played most often on the platform: according to the type of music and tastes of friends, with time is able to suggest new tunes that the user should like.

Lumi extends this idea to the whole web: once you get the plug-in (available for Firefox, Safari and Chrome), the service analyzes your browser history and creates connections with other quality content that should be relevant to you. The suggestions are then proposed in a nice and colorful page, almost magazine-style:

Of course, such a severe tracking can arise serious concerns regarding privacy. Luckily, Lumi’s terms of use are very clear: no type of personal information will be shared without your permission, and the history of your browser will remain invisible to anyone (even to Lumi’s team). In addition, you can always delete your account by removing all saved data.

At the time the application is self-funded and in beta with private access: you can obtain the plug-in by entering your email on site.

But in general, the approach of Lumi and similar applications is extremely interesting for more than one reason. As the web grows increasingly in size, it’s natural to find new ideas to retrieve interesting content: with the era of big data the analysis of information becomes a very complex art, and a search merely based on keywords can hardly cover everything you want to know (or anything that you would like to find out) about a topic.

Of course, in many ways the classic Google SERP is still more than enough, but it is also the residue of a fairly limited view of internet: and not surprisingly, with the Knowledge Graph move, Mountain View is trying to make its engine a bit more “intelligent”, capable of handling data with less ambiguity and to better interpret the requests made by the user.

Lumi and StumbleUpon, on the contrary, prove the importance of a qualitative, and not strictly quantitative, approach to web page retrieval: here is the information that somehow goes to the user, and not vice versa. Reworking this way the lesson of the social web – that is, just as it always has been, a lot of new information derives from the suggestions of other people – and trying to build it in “private”, carved in the heart of an algorithm


Comments are closed.