Design Cup Series Development Cup Series
As the first stage draws to an end for The Digital Run, I am going to take stock of this great new competition format. And a great competition it was, with veterans and newbies alike withstanding the challenges of a 13-week marathon and racking up some surprising results. In this article I will recap some of the highlights of the stage (mostly from the design side) and give my perspective on Digital Run strategy.
On the design side, Stage 1 saw some of TopCoder's top designers pitted against each other. The result: A new top-ranked designer, as aubergineanode's three wins over Pops in three head-to-head competitions helped him reach #1. aubergineanode threw himself into The Digital Run, submitting a record 24 design submissions over the 13 weeks. Rounding out the top three submitters were AleaActaEst, with 17, and Pops with 15 (though I should note that Pops finished his submissions in about 1/3 of the time that it took me to do mine! It's all about the power of reliability - more about this later).
On the development side, there were great performance from coders such as minhu and haozhangr, with 13 submissions each, and kr00tki, with 9 submissions We have also seen some new designers and developers achieve some great feats against some of the veterans.
So, is it a marathon or a sprint?
One of the common questions about The Digital Run has been what it would take to do well. Does it have to be a full-time, 13-week commitment? Or could occasional bursts of hard work and brilliance suffice? After the first stage, it's clear that the answer is "yes" - it can be both.
The Digital Run requires both long term planning and persistent focused execution, but also rewards sprinters, as competitors can accomplish significant gains in a short period of time.
Looking at the strategies designers have used, we can see these two extremes -- the steady week-in, week-out performance of designers like aubergineanode and AleaActaEst on one side, the occasional bursts of multiple submissions from Pops and real_vg on the other.
This is a mixed bag of strategies. But the bottom line is that even the occasional competitor who chooses his or her spots can do very well and make some great money. Look at the table at the end of this article to see how kyky earned $992.11 with just three weeks of submitting.
Planning is everything
Planning when and how to compete is critical, particularly because of the way The Digital Run format awards points. Points are shared among the competitors submitting the same component, with a 500-point pool divided among them. This means that if two competitors go after the same component -- and both pass the review -- then first place gets 325 point and second gets 175. The more competitors that go for the same component, the more that 500-point pie gets divided. This means that you could come up with 10 great designs and potentially only score 300 points for all that work.
Maximizing your gain:
So, how do you maximize your point booty? One strategy is to wait and watch who is interested in what designs. This can be done by
And remember: Watch the leader board all the time to see what the competitors are doing. Remember that great intelligence gives one intelligent alternatives.
Countering (or holding on to your spoils):
Sometimes you work hard on a component and submit a great design but find yourself getting only 75 of the 500 points because of so many submissions. Meanwhile, some other guy goes solo on an unpopular component and earns 500. With that stroke of bad luck, you have slipped down a spot or two in the standings.
In addition to picking your competitions carefully, you may also need to play defensively. In the scenario above, if you paid attention sooner and countered the other guy's move, you could have defended your spot on the leader board. Assuming you came in second and only the two of you competed, the other guy would only get 150 points more than you (his 325 for first place versus your 175 for second place), which is much better than the 500 points he could have gained with an uncontested submission.
Countering like this not only slows your opponent down, but also shows that you mean business. This was the case during Stage 1 when Pops made a leap and grabbed four component designs. aubergineanode countered him in this exact manner (on three submissions) to ensure that he could control Pops' point gains. (Incidentally, getting four components done at all, especially with high quality, is quite an accomplishment in its own right. Congratulations to Pops, aubergineanode, and real_vg, who all went for the big four!).
Ahh, many a coder has wondered about this little aspect of design/development. In case you haven't heard it before, let me be the first to tell you: Reliability is very important for The Digital Run!
Why? If your reliability is greater than 70%, you can potentially register for as many components as you wish and thus earn a LOT of points. This is a great option to have, allowing you to jump in and grab some easy designs as they arise. Consider what real_vg did in the last week of Stage 1 on the design side, when he went for four components, most with little competition. This is an awesome way to get some great point boosters and he could not have done that if his reliability rating was below 70%.
Bottom line? High reliability is a very important overall asset to have, especially for Digital Run strategizing. Plus, reliability of 80% and above translates into a bonus of 10%-20% on your prizes for winning components (though not on Digital Run prize money). Consider this: The winner of a $1,000 component would actually get $1,200 if their reliability is 100%!
I am new, I can't do this.
"Ahh, but there are too many reds and yellows," you might say. "I am new, no point in doing this." You could not be more wrong. Just look at the performances of some first-time designers such as skye85, librarian, or sindu (granted, maybe a special case, since he won the TCO in development). Everyone was new to design or development at some point -- the important thing is to start. That's the only way you will ever get better! Not to mention the Rookie of the Year prizes that are up for grabs.
And now for some stats, ladies and gentlemen:
Sometimes numbers tell the story better than words, so let's look at the stats:
We can think of The Digital Run money as being a bonus per component submitted. In such a case Pops, for example, has made $885 extra for each component submitted. Given that the top 1/3 of all submitters will get paid extra this is an amazing way to make money on top of what you could make with a component win! For designers, this translated to $137 as the minimum payout for a top third designer (dplass in the table) for each component submitted. So you can make money when you actually do not even win a component!
Challenges for Stage 2
I think that haozhangr's quote is very appropriate here: "Do or do not. There is no try."
There are a couple challenges that I would like to put forth to "the doers:"
Stage 1 of The Digital Run was everything I had hoped for: a great challenge, a great opportunity to make some money, great drama when trying to see who is doing what (in fact, as I write this I am not sure if I will lose my second slot to Pops!), and a great experience in all (apart from a couple sleepless nights, but hey - who needs sleep?).
I want to thank TopCoder for creating such an awesome contest, and I hope to see you all competing in Stage 2.
*This is still not finalized as the designer has outstanding designs
(1) Note that all ratios are rounded up
(2) We can think of The Digital Run money as being a bonus per component submitted. In such a case Pops, for example, made $1,021 extra for each component submitted.
(3) These numbers are still tentative for some members and might result in changes to this table (including changes to position or actual placement in the top 1/3, which this table represents).
*This is still not finalized as the developers have outstanding components.
(1) These numbers are still tentative for some members and might result in changes to this table (including changes to position or actual placement in the top 1/3, which this table represents).