CVRcade Vision Point – Physical Equalization™ – Gaming the System
by Frank Garcia
April 15, 2019
In my first vision post on Physical Equalization™, we talked about the benefits of PE, but no system is perfect, so let’s talk about PE in that light.
We learn from Gödel, Escher, Bach by Douglas Hofstadter (1979) that any system once constructed can be gamed. The best you can hope for then is to make it very difficult to cheat. Currently, most cycling esports systems are very easy to cheat. One of the primary design concerns of CVRcade was to make cheating difficult. We did this on two fronts, first through Physical Equalization and second through data analytics.
Physical Equalization, or PE for short, calibrates every rider and their equipment against every other rider and their equipment and normalizes that performance. Counter-intuitively, and very much unlike other systems, if someone cheats in some manner, while that will be effective in the first race in which they did it, it would very much hurt them in the long-term. This is because their artificial level of performance would become their new “normal” (their new expected level of performance).
To gain an advantage over time, they would have to increase their cheating each day they raced in Cade. This could of course still be done, and so that is where the second part of the system kicks in, data analytics.
Each day we analyze the races and check for “aberrations” in the data. Continually increasing performance that does not model human physical activity is easily identified. Keep in mind that we have a tremendous amount of telemetry on which to base analysis.
Our first approach is not to go after riders but to improve the system. Whenever possible, we improve the system to guard against “anomalies.” So, for the most part, if there is an issue we consider it to be our issue and our responsibility to address it systematically for the future.
Now, as stated above, and because frankly there are some bad actors in the world, it is still possible to cheat. Rest assured that when we identify people who are cheating, they will be promptly, permanently, and unceremoniously booted from the system.
A more subtle approach – Sandbagging
Some have talked about using sandbagging as a device on Cade to be able to do well over time. And, if only PE were used, it would indeed be the best way to attempt to game CVRcade. But through the use of analytics, we are able to address this (to be clear this is included in some of the tweaks that are arriving shortly).
As above whenever we see continuous improvement over time that does not reflect typical human behavior, we first start by applying adjustments to a person’s normalization. This means that if they are cheating it becomes counterproductive. If not, we can identify that and make the appropriate adjustments. We won’t share all the details on how this is done but know that it is being done.
Perhaps more importantly, sandbagging is mitigated by performance and competition. If you sandbag yourself out of competition, then, for the most part, it can just be ignored (we will still identify and kick out people who cheat). If you are being competitive, then you will quickly have to increase your performance at each race as you set new normals, doing this will also highlight you to the analytics engine.
One can see how this makes leagues and series, or just having people race multiple times, an important aspect of creating fair and trusted competitions especially at the big, high-prize purse events, that require you to participate in a league and thus a series of events to reach.
The Wild West – Newbies
New people coming into Cade present a challenge to PE, I will do more posts in the future on this subject. But to address it to a large degree here know that the simple answer is that it will be addressed through real-time adjustments when in ranked events that will make it very difficult for newbies to “podium” (while still being able to ride with the bunch) until we have sufficient data on them. This is a change that will be coming shortly.
The Current State
There are a number of tweaks coming to PE. You have already seen some. We have talked about several important ones above. There are others. For instance, an important but simple one is that we are addressing inappropriate max HR values (caused by clearly inappropriate transient spikes put out by an HR monitor).
ALL The people riding in beta have helped tremendously. Your efforts and approaches along with lots of analytics are allowing us to put the finishing touches on equalization in the coming few weeks. We are most appreciative. Thank you for all your help.
Physical Equalization in combination with data analytics is an amazing system. Get on Cade, give it a go, and give us your feedback so we can make it even better.