Scoring system balancing

[devskill]

Today we have deployed a major change in the scoring system. This is part of our recent reshaping of Dev Skill profile. We have removed many useless components to make it more meaningful and useful based on current need.

Due to change in the scoring policy we have cleared the scoring progress history (hopefully this is the last time we have to do it) so that it do not show a downward graph in your profile. Downward graph can indicate negativity as it proves you have been doing bad over time. So we had to clear this up.

From now on Dev Skill scoring system will be more dynamic and predictable. You can calculate your score on your own based on the forumla. Here is the full detail of the new scoring system can be found in our FAQ page here: http://devskill.com/faq

Though based on the new score balancing your score may have dropped significantly, but it is more realistic and over time you will find doing activity in Dev Skill will generate score in more predictable way. Solving problem, participating in contest, and even submitting problem to us can now give you proper score.

We have reduced the dominance of UVA and balanced it out to be only one of the factors. From now on, Dev Skill's own score will be the leading score generator. The other 3rd parties will contribute in the score, but will not be a dominating factor. Overall we have balanced the scores in such a way that no single factor becomes too dominating.

[froghramar]

I want to give some feedback based on your scoring system.

1. Max Penalty: In a contest max penalty is 95%. It can demotivate contestants to solve a hard problem at the very last moment after some unsuccessful submission. I would have preferred 60% max penalty.
2. UVA Score: Won't Round(((UVASolveCount * 100.0 / 1500) * 1000 ) /100.0) give the same result as Round(UVASolveCount / 1.50)? If it is then, second option is more understandable.
3. Stackoverflow Score: I'm not happy with the score from stackoverflow reputation. I mean, 1000 reputation for just score of 1? Though I am too young to say like this, but I think earning 1000 reputation isn't as easy as solving a beginner problem and both gives you same score.

Hoping for better success of devskill and BD programmers.

[devskill]

Thanks for the feedback.
1) We will reconsider max penalty. It may not be 60% but we will figure out something in between that helps.
2) The equation is already updated, thanks for pointing it out. Actually we wanted to explain from where 1.5 comes from. But anyway, it is not making that much sense and we have updated it.
3) For stackoverflow, we have mostly focused on Medals instead of reputation. Reputations generate without you actually doing anything based on your posts, and thus sometimes grow exponentially. For example, it is possible to gain 70,000 reputation. Initially it is hard to get reputation but for old users, it grows rapidly. if we do not limit the score in reputation, it can unbalance the score gain. But we agree, getting 1000 reputation to 1 is a bit unfair. We will think about it and may come up with a range based calculation.

Thanks

[devskill]

Wanted to let you know that,

we have already made some adjustments.
1) Now the max penalty is 70%
3) Formula was adjusted to consider Reputation * 0.01 instead of Reputation * 0.001

[devskill]

A new patch was added in github scoring: new formula is : github score = (FollowersCount * 2) + Round(CommitCount * 0.1) + PublicRepositoryCount * 2;