Daniel Lurie, Editor-in-Chief of DataArt.
Twenty years of DataArt was developing IT solutions for Clients from the sphere of finance, health, tourism and other serious industries. We never released our own products – we feared a possible conflict of interests. But recently we launched our own Skillotron gaming platform. It all started with the fact that the DataArt appeared two seemingly unrelated problems.
The first – the assessment of the professional level of colleagues. Over the past ten years, we have grown from 400 people to 2500, and people continue to come to us, DataArt already operates in 10 countries and 20 cities around the world. As with any organization, we, of course, have a qualification evaluation system for hiring. It is more difficult to assess those who have worked with us for years and all the time. Partly for this, technological communities within the DataArt arrange "self-checks" – voluntary exams or olympiads, where everyone can adequately assess and understand what else he needs to pull up, in which direction to develop. This process is quite laborious – it took a lot of time and effort of the most experienced colleagues to come up with tasks, organize a self-check and check the results. The developers wondered how to simplify life for themselves and the rest.
The second problem was that the 20th anniversary of DataArt was approaching, and it was necessary to create a new site. We wanted to create something non-standard that could distinguish it from other corporate sites and tell us something about us, for example, a test game dedicated to IT. The site team decided to discuss the idea with the developers. It turned out that they had long been thinking about something similar.
The idea of a million hours
The developers wanted to do a test based on the "skill tree". Such a thing would allow us to evaluate not just the abstract level of Java ownership, but separately to assess how much a person knows Core libraries, separately – Ui libraries and so on. The site team thought about a lighter genre, casual game – so that every visitor could spend a couple of minutes on it, have fun and send a link to friends. As a result, the developers managed to prove that a simple test is uninteresting for engineers, although they agreed that the test should not be too serious.
The development of the first public version took six months. The game certainly could be done faster, be DataArt a grocery company. Alas, the main forces at us always go to client projects, so it was almost impossible to build a strong team and protect it from the constant encroachments of commercial projects.
In order to come up with a format for the game, we studied similar systems – from serious platforms for certification to Applications for preparation for exams in a driving school and language learning. As a result, we came to such rules. Each game session consists of 10 questions on one of the technologies. In the event that the player answers correctly at least 8 out of 10, he goes to the next level with more complicated tasks. For correct answers, and when going to each next level points are awarded, and during the game participants get bonuses.
The first release of the first version of Skillotron. There are just over a thousand questions divided into 16 categories and several dozen subcategories, from programming languages to project management technologies and IT history.
Skillotron was created as a system that the community should support – those to whom it is Need and interesting. Participants can not only answer other people's questions, but also ask their own. They also moderate all content on the site. The number and quality of the questions are in the hands of the community, and their complexity counts the algorithm.
About half of DataArt developers have already played in Skillotron, each an average of 4 times. The return of users is 76%, and the average session time is 9 minutes 20 seconds, quite a good indicator.
We attract third-party users to the project gradually, without massive advertising. We will look at the statistics, receive a feedback, improve what needs to be improved and move on. It is important not to interfere with the community, you need to follow people. The main goal of the first stage is to debug all the mechanisms that will allow players to convert to the authors of questions. They are the most valuable.
As with any project built around the community, we really do not know if Skillotron will take off, and if it takes off, which way will it go. Nevertheless, we already have ideas on how to develop the platform in the future.
Skillotron can be turned into a tool that allows you to not only evaluate your current level, but also increase it – in case of incorrect answers, it can suggest that Worth reading to improve their results. We are thinking of using a platform for conducting remote interviews: the candidate rips the screen and answers questions, on the air commenting on his decisions. Skillotron can be useful in matching customers' requests and our assessments of the professional skills of colleagues – that is, for example, under the middle developer, both we and customers understand the same thing.
Skillotron is now a classic test : For any question you need to choose the right answers. Soon we will start adding questions, the answer to which should be the code written by the user. This code will be executed in a container and checked by tests. Such tasks and compose more interesting, and to answer them more difficult.
In the process of growth and improvements Skillotron it is important not to lose the principles on which the platform was built. It must grow and develop organically. We still keep the program code, but if everything goes well, then it is quite possible that we will publish it in open source.