我只想搜索
I Just Want to Search

原始链接: https://www.0xsid.com/blog/just-want-to-search

现代搜索已从精准检索工具演变为由“相关性”和算法推荐驱动的不可预测系统。曾经,“Google-fu”(指运用特定搜索指令和引号的技巧)是一项核心能力,而如今,各平台经常忽略这些指令,优先展示其指标认为用户“应该”看到的内容,而非用户实际请求的内容。 这种衰退在网页浏览器、电商平台、电子邮件服务和视频平台中表现得尤为明显。用户失去了控制权,被迫应对 SEO 垃圾内容、被无视的搜索指令以及隐藏的过滤器。虽然转向模糊语义匹配可能是为了有效处理海量数据,但这却以牺牲那些仅需要字面结果的资深用户为代价。 作者认为,企业将整体参与度置于实用性之上,致使那些清楚自己需求的用户失去了“手动模式”。解决方案很简单:各平台应提供一个“仅字面匹配”的开关,绕过预测算法,实现可靠、精确的检索。在人工智能驱动内容策展的时代,高性能、用户可控且优先于算法猜测的搜索工具,存在着清晰且未被开发的市场需求。

Hacker News 上的一场讨论探讨了搜索引擎质量的下降,这是由针对 SEO 垃圾信息和低质量内容的无休止对抗所导致的。用户“gtowey”认为,自动化系统在这一“猫鼠游戏”中注定会失败,因为垃圾信息发布者总是会比算法领先一步。 所提出的解决方案是转向人工编辑控制,这将打破操纵循环。尽管谷歌等科技巨头因成本高昂而避免采用这种模式,但作者建议采取类似维基百科的去中心化、志愿者审核模式。虽然并非万无一失,但维基百科的模式已被证明在应对复杂的攻击时具有极强的韧性,为提供更精选、更值得信赖的搜索体验提供了潜在蓝图。
相关文章

原文

Learning to search was a key skill growing up as a teen in the 2000s. How to use accurate keywords, quotes, the entire lot of search operators. Google-fu was a hard earned skill (yes, I read a book) and helped me figure out a surprising amount of life.

A vintage instructional guide showing a 1900s-era AltaVista search engine screenshot, instructions on basic internet searching, and a list of defunct and classic search engine URLs. How to Do Just About Everything on a Computer, 2000

Search these days though is a bit of a dumpster fire. It has shifted from just retrieval towards relevance and recommendation and it sucks. I’m not just talking about web search, which has oscillated between “help me find the link for this product” to “somewhat useful if you can ignore the SEO spammers” [1]. No, I’m lamenting how poor search has gotten inside emails, online marketplaces, and messaging apps. I’m lamenting that the user has progressively lost control over what constitutes a search.

My beloved AV receiver died a few weeks back. No biggie (even though it’s EOL), let’s see if we can find the same model being sold for parts so we can try and replace some board components.

A Facebook Marketplace search for the home theater receiver query '

I’m betting you’ll get inundated with listings of the same brand, not the model you’re looking for, even if you put in the exact model number in quotes. It’s not limited to obscure AV equipment. Searching for any item that’s not in oversupply means the quotes get basically ignored while the search branches into anything that might make sense.

Searching on YouTube means scrolling over two entire rows of Shorts,"People also watched", "Explore more", with your actual results somewhere in between [1]. Time related filters have gotten lost too: not being able to choose a date range (unless you manually enter date operators into the search field), not able to sort by date at all, or just videos from the last hour for breaking news.

Gmail is a whole other story. An exact invoice number or name pasted into the search bar in quotes sometimes returns nothing. Is it a genuine tech issue, or did Gmail quietly switch me back to “Most relevant” again?

And yes, I know. I can’t fathom the scale these services run at. Full text matching across billions of inboxes is computationally expensive. You’re indexing billions of emails and attachments, keeping those indexes fresh, and trying to return an answer in milliseconds. At some point, I can see someone looking at all of that and thinking: what if we just show people the results we think they want?

The en-shittification angle is maybe too cynical, too simplistic. Maybe this really is the natural progression of trying to make search better. At some point, relevance became more useful than literal matching for most people and the defaults followed the metrics. The problem is that there’s increasingly no mode for people who know exactly what they're looking for and want just that.

So what exactly is the point of writing this? I want the random PM looking for ideas to know that I’d be very happy to pay for better search on platforms that help me get what I need instead of optimizing a metric. Give me a “literal match only” toggle. Kill the fuzzy suggestions, ignore the semantic guessing, and just run a dumb, reliable grep across my data, boss.


[1] I got so annoyed I just spun up a version of my decluttering app just for YouTube.


If you've reached this far, thank you for reading! :)

I thought retiring in my mid 30s after a few exits would be fun but I've just been bored and a bit undersocialized without morning Slacks and emails to wake up to. If you’re building something interesting and could use an extra set of hands to ship, or just want to say hi, feel free to reach out. My inbox is open.

联系我们 contact @ memedata.com