What estimator did you use to score the result?
What estimator did you use to score the result? Did you verify the scoring result across different interfaces? I downloaded your .swf files and load them in sabaki and the scoring result is completely different. From the result of your article I believe that you used an outdated estimator where the dead stones have to be hand picked or if you wrote it yourselves then I believe you completely misunderstood how the Go scoring system works.
I think you may find our explanation in https://boardgames.stackexchange.com/a/58133/39427 informative.
I'm going to leave this issue open for now as a kind of FAQ, and close once we have an updated version of the paper on arXiv that clarifies this.