The Tech Machine: Proposal
Goal Summary
- a better way to discover new, up-and-coming apps/products
- focus initially on app discovery, with end goal of general tech product discovery
- ultimately, an easier way to search a space within the tech industry for pre-existing products (useful for benchmarking)
Original Targeted Problem
- current benchmarking process (that we underwent for this 210 project) is flawed
- consists of several hit-or-miss Google searches highly dependent on the user’s ability to form appropriate queries -- and even if you successfully discover one or two standards in the desired space (be it an idea or fully-formed product), there is no obvious pathway for the discovery of similar standards
Product Vision
What if there was a way to more quickly perceive the competitive space for a new product idea of yours? This cannot be accomplished efficiently with the current news/feed aggregators like CNET or The Huffington Post or Google News - although these sites may contain the information we want, it is usually scattered throughout the sites and difficult to find (“Related Articles” sections and article tags tend to disappoint). We would like to index and aggregate the information from these types of tech feeds and blogs in an intelligent way - specialized for the online tech community, organized by noun entities (via some basic Named-Entity Recognition?).
Inspiration: The Hype Machine (or HypeM)
A web service that aggregates leading critical voices in the online music community. An excellent way of discovering new songs and artists, typically independent of the mainstream music cycle. It pulls content from 800+ mp3 music blogs and creates a sort of trend map of what music is being ‘hyped up’ in real-time within the indie community.
Our job is to sort that [i.e., how overwhelming music blog posts are in their number, size and frequency] out and make it more accessible, easier and more user-friendly.
- Anthony Volodkin, HypeM founder
Site mechanics:
- HypeM crawls a variety of mp3 blogs for new posts. If a post contains mp3 links and embedded media (from sources such as Soundcloud), HypeM will add this link to its database and display it on its front page under “Latest” blogged music.
- Tracks are stored and sorted by title and artist, and links to all the blogs that have posted a particular track are aggregated underneath a single track listing.
- Users can then search for an artist or track, play the song directly on the site, and “heart” the song (which will add it to the user’s favorites playlist). Users can also access other sitewide lists similar to “Latest” such as “Popular,” “Featured,” and “Genres.”
- The “Popular” list in particular is interesting because it breaks down into several categories: popular now, this week, in past weeks, on Twitter, by artist, etc. This list is dependent on user-provided “hearts” (i.e., favoriting) and provides finely-tuned metrics on trends happening within the indie music community over time [also, see section below on HypeM’s Music Blog Zeitgeist].
Advanced Inspiration: HypeM’s Music Blog Zeitgeist
- Now that these mp3 blogs have been so thoroughly indexed, at the end of every year, HypeM compiles several “best of” lists for the year - for artists, albums, and tracks. The “Top Tracks” list is based on song favoriting activity from the user community over the past year.
Our Product Idea: The Hype Machine for Tech Products
- An Android app that aggregates the leading critical voices in the online technology community
- Provide a way for users to: 1) quickly and simply discover up-and-coming tech products, and 2) follow trends in the popularity and community enthusiasm surrounding a product over time
- Initially, we will focus on discovery and trends for up-and-coming Android apps
- Our product will crawl a selection of tech news aggregators and blogs for new posts on up-and-coming products, and these products with be displayed by name on our main page in real-time
- Interface will be very simple and easy to glance over - a product name, list of source posts, and some sort of “pulse metric” (e.g., # of source sites talking about this right now [although using this might lead to too many ‘mainstream’ products being displayed], # of users who have ‘watched’ or ‘liked’ this product, etc.) should be all we need
- App will not display extraneous product information (that’s why we link to the source sites) or mere article headlines (like an RSS feed) - we need to extract a product name from the articles
- App will also have a “Popular” ranked listing based on our “pulse metric”
- App will also contain search functionality, useful in allowing users to cultivate their own product watch lists
- Extension? Products can be tagged and linked to other similar products by users in order to create a more easily-searchable network of products - this functionality will bring us closer to the original “benchmarking tool” goal discussed above
Two observed methodologies: Crowd-sourced vs Personalized recommendations
- i.e., HypeM model [crowd] vs Pandora Internet Radio model [personalized]
- Crowd-sourced: geared towards simplified discovery/industry pulse & trend readings
- Personalized: geared towards providing a suggestions & decision-making engine
- Our product idea: crowd-sourced model
Competitors
- Kinetik/HIP [app]
- How it works: app recommendations platform organized around a newsfeed, a friendfeed, and a dealsfeed; newsfeed pulls reviews from tech blogs and the App Store; lets users favorite apps they already have installed to share this info with friends [crowd-sourced/personalized hybrid]
- Shortcomings/Distinctions: very poor (average: 2/5 stars) App Store ratings, newsfeed implementation does not distinguish between positive and negative reviews (informative article found here)
- Discovr Apps [app]
- How it works: creates an interactive map of recommended apps using any app at a starting point [personalized] (informative article found here)
- Shortcomings/Distinctions: iOS only, user needs to have a product in mind to start with so it’s not based on trending popularity/enthusiasm
- Best Apps Market for Android [app]
- How it works: advertised as an alternative to the Android Market similar to editorialized iPhone App Store lists; reviews, categorization, and trait tagging done by hand by “Android experts”; offers personalized recommendations based on installed apps [crowd-sourced/personalized hybrid]
- Shortcomings/Distinctions: unclear who these “experts” are, and thus the quality of listings and recommendations provided is questionable
- Hypejar (Gadgets section, in particular) [web]
- How it works: a Wikipedia-style repository for upcoming product launches, including movies, music, videogames, tv series, gadgets, events, books, etc. [crowd-sourced]
- Shortcomings/Distinctions: products only appear on site if a user creates an entry for it (similar to Wikipedia), not for apps (informative article found here)
Technical challenges
- Developing a highly-intuitive UI for this kind of Android application
- Our model (HypeM) is a web service, but we want to build an app
- Music and apps have inherently different methods of representation, music perhaps being the more intuitive one (music charts are a solid model that HypeM could emulate, but the App Store’s Top Charts may not be the best model for us)
- UI is critical for this kind of app - it needs to be useful but streamlined
- How to index/extract apps from various tech blogs, how to perform accurate, efficient information extraction
- HypeM has not publicly discussed its method for indexing, in real time, 800+ music blogs
- Nontrivial technical challenge without previous experience or guidance
- Choosing our set of tech blogs/feed aggregators to index from initially
- Need to start with a small set of sources that will still give us a representative sample of online tech community
- Use Technorati for blog suggestions?
- Podcasts are also interesting sources of tech product discussion..? [stretch]
Tie-in with SAP project theme
- Honestly, a bit of a struggle to tie-in to user identity
- User accounts available to allow users to express enthusiasm for apps or products - social in that it allows for curation of a site identity that expresses your interests within the tech community
- Other ideas?