Why I, a Non-Engineer, Built the One-of-a-Kind "Steam Evaluation Lab"

The "Sediment" Problem in Steam Reviews

I am not a professional engineer. But as a gamer, and as the operator of NK Labs, I couldn't ignore this "information gap". Many sites focus on whether there are "fake reviews" (shills). But I thought it was more important to look at "how evaluations have changed over time (sediments)". Old evaluations become "noise" that doesn't tell us the current state of the game.

Building with AI (Antigravity)

"If I can't write code, I'll borrow the power of AI." Deciding this, I started development with Google Antigravity, the latest AI partner. As a non-engineer, my "ideal" was shaped by AI. After many iterations, the unique calculation method "NK Score" was born.

NK Score: A Formula for True Value

This lab uses the following formula to reveal the "current" true value. Here, $R_i$ is each review's rating, and $w_i$ is a "weight" based on posting time and playtime. We value recent evaluations highly while filtering out shills and noise to extract only the "pure evaluation". Specifically, we look at:

• Freshness of reviews: Newer reviews are rated higher
• Playtime reality: Extremely short or long times are corrected
• Account status: Newly created accounts are treated carefully
• Content: AI reads the text to check if it speaks specifically

"Before buying a game on Steam, check the 'sediment' at this lab first." I aim for a world where this becomes the norm. I don't want to make anything half-hearted. I will continue to combine the power of AI with a gamer's perspective to provide the sharpest and most honest appraisals anywhere. NK Labs will support you with all our might so you don't say "I regret buying this..."!