當搜尋能分辨投機伎倆與專業本領
When search can tell a trick from a craft
搜尋系統開始理解網頁內容，而非只計算字詞數量，因此能分辨真正説明業務的頁面與刻意迎合排名的頁面。這或可讓小企業把原本用於操弄搜尋排名的時間，重新投入自身業務及互相轉介客源。排名訊號、檢索流程及提示詞等設計，則由編寫程式的開發者逐步決定。
For ten years the small firms on every street wrote for a machine. Keywords in the right density. Backlinks bought by the hundred. Eight hundred words around a question nobody asked.
It worked, mostly. The machine could count, and counting was easy to fool.
That is changing. The systems that now read the web on our behalf do not count words. They read them. They can tell a page that explains what a business does from a page that tries to look like one. Plain truth starts to rank. The trick is walked past.
Think about the energy that went into the trick. Every hour a small firm spent gaming a search page was an hour not spent on its own craft. Multiply that by every small business on earth. It is a large amount of human effort, spent on a fight that produced nothing.
When the fight stops paying, that effort comes free. Firms can say what they do and for whom, and then go back to doing it. Two companies in the same street can send each other work instead of outbidding each other for the same spot.
None of this happens on its own. The judgement that separates the wheat from the chaff is built, line by line, by people who write code. Ranking signals, retrieval pipelines, the prompt that decides which source an assistant trusts. Those are design choices, and developers make them every day.
So the question lands with you.
You are writing the code that decides what gets found. What do you want it to find?
來源：DEV Community · dev.to