Lifehacker performed an interesting survey recently, asking readers if Google’s search results have become less useful lately. This follows a handful of mildly negative articles around the web questioning Google’s future.
- 34% selected “Absolutely. The spammers have gained a significant threshold.”
- 44% selected “Kind of/sort of, but it’s still the best way to get at the good stuff.”
The article is entitled “Over 77 Percent of Lifehacker Readers Say Google’s Search Results are Less Useful Lately”, which feels somewhat misleading. “kind of/sort of” isn’t exactly a conclusively negative response.
The survey itself is a bit awkward from a research perspective. What does lately mean? A few weeks? A few years? Statistically speaking, it would be difficult for a human to recognize a small decline in a short period of time (like say a week or two). If the search results had become drastically worse, maybe there would be enough evidence to legitimately observe a decline.
That said, unfortunately this involves a case of perception and the Internet. We expect Google to work well. If we expect it to work well, good search results don’t stand out in our minds. Instead, bad search results will stand out. In many situations, we won’t notice small improvements in ranking quality, but we’re very likely to notice terrible results.
thinking about solutions
From Google’s perspective, this might be treated as a cost/benefit problem. The benefit of an abnormally good search result is (relatively) small compared to the existing benefit of good search results. However, the cost of an abnormally bad search result compared to typical bad results is huge! Also, the amount of space to perform worse is enormous: Even ignoring SEO link farms for a moment, imagine searching about how to diagnose fluorescent lighting (from the US) and get a result on the culture of Tokyo.
All that said, as society derives more income from Internet business (including advertising), SEO becomes more important for income. And business practices can sometimes be unethical to increase profit. That’s not to say that all SEO is harmful; it’s certainly a good idea to ensure that you use appropriate keywords, meta tags, etc. And it’s usually good to be linked by people.
However, link farms are inevitable. Google is the top search engine so it will be the most heavily targeted engine for SEO. In some sense, it’s like viruses; they’re more common on Windows than Mac because Windows is a more common platform.
The Lifehacker article has a quote about an example search: “great headphones”. Search engines don’t really work like that. You don’t specify a goal; you specify keywords. Is a review likely to have “great headphones” in the title? Or often in the body? Probably not. Even the language of reviews is different: A reviewer might describe headphones as having excellent fidelity or being a good buy, but “great headphones” is a little to appear more than once in most articles. The commenters provide some excellent advice: This kind of query works much better when limited to discussion results.
That said, I’d like to point out that there are many different kinds of queries, and they’re affected by SEO link farms differently. Sometimes I use Google because I can’t remember a url or I don’t want to be bothered to type it all (or I want to rely on Google to fix any typos). For example, the UD course search page is some strange url and isn’t linked from the main UD page (you need to follow a couple links and the formatting is cluttered). But it’s the #1 result on Google. I might do something similar if I can’t remember someone’s username and I want to see their homepage. Or I might search for “wikipedia sublimation” if I can’t remember what sublimation means. These kinds of queries aren’t really affected by SEO link farms.
A slightly different query is when you know something exists on the Internet, but not where. For example, I might see a solid-state hard drive on sale and I want to know if it’s worth it. I’d suspect that I’ll find results over at AnandTech, but sometimes smaller sites carry good reviews also. In this case I know it exists and have a good idea where, but I’d let Google decide and would search using the brand/model name.
Then there are things you don’t know exist. Research articles can be like that. If there’s a review article on domain adaptation for part of speech tagging, you want it. But it might not exist, so you either find such an article or you try a few queries before deciding that it doesn’t exist (or it too underground to find).
In any case, the field recognizes that there are many types of queries (wikipedia gives some) and the problems and solutions with each type can be different. In this case, link farms really don’t affect certain queries, and sometimes it’s debatable whether queries ending at link farms are properly formulated or not.
I don’t think Google is going downhill quite yet. Their product failures are the result of public scientific experimentation and they’ve said before that they have the money to try things out.
SEO would be a problem for any top search engine, though I think Google might be able identify link farms automatically and push them down the rankings.