New Math Could Stop Big Retail Sites From Burying Small Brands

Every time you open an online store, something behind the scenes picks which products you see first. That system usually chases maximum revenue, which means the popular stuff wins again and again. A handful of bestsellers soak up most of the purchases, and everything else quietly disappears. This is why the same five brands show up on your homepage no matter what you search for. A paper posted on arXiv by researchers Xiaotie Deng, Hanyu Li and Chenghua Liu tackles that problem head-on.

Their idea is called balanced market share. In plain terms: if a brand gets any sales at all, it has to get a meaningful slice compared to the top seller. No brand gets buried in crumbs. At the same time, the store still can't show you everything — each shopper sees a limited set of products, maybe just a few at a time. So the store has to rotate its full catalog across different customers, showing each person a small curated menu while keeping the overall picture fair. That combination is genuinely hard to solve, and the authors prove it: even the simple version of the problem is computationally brutal, the kind of puzzle computers can't crack exactly.

So they did the practical thing. They built an algorithm that gets very close to the best possible revenue — within any margin you choose — while strictly keeping the fairness rule and the small display limit intact. It also stays simple enough to actually run, rather than requiring an impossibly large calculation.

What does this mean for you? If you sell anything online, it suggests a future where the algorithm doesn't decide you're invisible before you've had a chance. If you're a shopper, it means more variety and less of the same sponsored wall. The catch: this is pure mathematics right now, with no retailer using it. Deciding the exact best answer stays out of reach, so stores would be choosing 'very good' instead of 'perfect' — and only if a platform decides fairness is worth the trade.