Today our startup went version 1.4. As HN somehow accompanied the growth of our product, we would like to place the retrospective part of the birthday party as a case study here.<p>Rewinding a few months:
Our initial idea was to create a place where you can watch random snippets from any place in the world in realtime.
We decided to call it Whapee, an acronym for "What happens here".<p>When we were thinking about the features it should have, we had a growing list with a big trend to be infinite.
It should be independent, visual, free, respecting privacy although using LBS and operable without editors.
As the Android OS was hardly diversified when we started, we choose iPhones as the eyes of our project.<p>Our service had to find users taking and directly uploading images with their mobiles.
These should be geotagged automatically, commented by the users and stored for a limited lifetime on our server.
Our database should be searchable for locations or comments.
All uploads should be anonymous.<p>A website for watching images, an app for taking and watching them, plus an around me feature.
Main difference to all the camera apps and LBS locators in the wild would be the kind of non-vain pinboard that only accepts live moments.
You leave a snippet of what you actually see, or you watch what others look at in this moment.
Nobody knows you (sometimes one might guess), nobody collects albums and the whole process is completely transient.<p>We too didn’t have any idea which direction this all might take.
Kind of news, partyscout, hideout, but surely a mirror about life.<p>Our initial idea to use Google's geocoding capabilities were stopped abruptly when we inspected their terms of use. Every time you geocode an address or reverse geocode a location, you have to show a Google map. Everything else violates their terms. Further there are limitations of the API calls. At no cost Google allows infinite API calls.
Understandable, but this neither kept us independent nor could we find any sense in showing Google Maps whenever a user takes a photo.<p>With the discovery of the OpenStreetMap project we found a valuable replacement to Google Maps. As our decision was to have a clickable map plus a fully searchable geocoder, OSM had been the perfect choice. There are still black holes in the outback, but they are filled quickly and nobody needs an street number accuracy there.<p>OSM limits the API requests to its geocoder too, but allows to setup your own with their data. That said, our project had to extend.
Weeks went by when we had to setup two concurrent servers with geocoders that should give themselves backup.
Weeks, as the indexing of the data alone takes some days. Plus some weeks to learn, that Amazon's web services aren't suitable for some hundred gigabytes in a PostgreSQL database.<p>Before launch, we introduced a lot of abuse preventers. Shielded our CPU consumptive geocoder, added spam filters and quick view tables to immediately remove material that’s inappropriate.
All had been prepared for a furious start, reliable and hardened. Looking at these preparations today makes them look exaggerated. There never has been anything, that had to be pulled of the board.
From a professional view it’s good to have them, but they delayed the start again.<p>A thousands test images later we went online. While thinking in months before we could reach the break even, we expected a good start followed by a vastly declining interest. Instead we found us widely ignored by the press and the number of users that check out every new app has been lower than estimated.<p>Version 1.0 perhaps was too ambitious in being off from all known social networks. As we set our focus on privacy, we decided to stay away from every account-based link to our images.
For an app that is used to upload pics and messages beneath a bunch of similar products, this is like committing suicide. At least it felt like this so we had to adjust our policy.
We decided to add Twitter as a linker to our images if the user liked to. We left out Facebook by intention.<p>We were thinking of bad months before and all the funds had been prepared for this. So we started the phase „Didn’t help but improved experience“.
The daily flood of new apps killed every chance of a self-runner. Working out all the points that are unique doesn’t help when nobody listens. No domino effect from friends to friends took place.<p>Localization. As we tried to cover the whole world, we had to come towards it. A good online translation service gave us the ability to discuss the translation down to the length of button captions. We learned, that Russian is one of the most luxurious languages regarding the character count. We were overwhelmed by eligible questions from our translators we had to answer quickly. One of them was always at work while we had the lights off. Nevertheless, this little Babel adventure did not send us new users automatically.<p>With todays birthday, we are still away from the number of users we need for perpetual uploads.
But hey, we luckily have been prepared for this, so the situation didn’t went uncomfortable.
We knew that our service will be early and we have to wait for a growing sense in privacy and LBS.
There is still the risk, that our idea doesn’t attract too much people or there is no minor majority willing to post snapshots without being credited. But watching the view from a skyscraper in Osaka while having breakfast in a forestal flat at the same time, is worth all the time dedicated.<p>The first public image ever made for Whapee has been the view into a court from China. In return, we integrated Sina Weibo today.
Still a lot to improve.