To return to the original question, in my view there's nothing wrong with google using synonyms as part of the process of determining their search results. In some cases it might well improve the results.

But they should only use LEGITIMATE synonyms.

The problem is that they've apparently stretched the process to include entities that aren't true synonyms. This has caused the algorithm to sometimes give high rankings to pages that don't have the information that the searcher is looking for. It wastes a person's time when they go to these pages and can't find the needed information. It's a bad user experience.

In other words google has taken the use of synonyms too far, to the point of mis-using them. That's why the quality of their search results has gotten worse.

Apologies for getting a bit self-involved there...

The problem is that they've apparently stretched the process to include entities that aren't true synonyms. This has caused the algorithm to sometimes give high rankings to pages that don't have the information that the searcher is looking for

"Sometimes" as in "rarely," or "fairly often," or "moderately often," or "very often," or "nearly all the time"? There's a big difference between each definition of "sometimes."

In any case, search (like so many other things) is a work in progress. That's as true today as it was back in 1993 or 1994, when I first started using search engines.

Shopping for shoes but Google would much rather try to sell you some socks via the listings presented -- Happens all the time.

Can anyone confirm that when searching shoes, that the ads or shopping blocks show shoes? Or do they show socks instead? I know. I know. It's a rhetorical question...

I just searched Google for shoes, and there wasn't a sock in sight.

Maybe google's personalized search can determine some people's dressing habits.

<irritated digression>
I'd be everlastingly happy if they could just rid themselves of the notion that "" (the empty string) is synonymous with string-that-has-content. I spend a fair amount of time looking up obscure illustrators by name, like "John Smith illustrator" and every time I get a flurry of results offering to sell me Guess Which Commerical Product, with results that proudly and avowedly do not contain the words John and Smith. (Fortunately there is no Major Commercial Product called Artist, so I can fall back on that--and hope that at least once in John Smith's career, he did something that would cause the word "artist" to be associated with his name.)

The irony is that I do all of this from the Advanced Search window, meaning that if I wanted to search in the form "result must contain at least one of the words John or Smith or illustrator" I could readily do so in the first place. Instead I'm entering the search query into the box that clearly and unambiguously says "all of these words". (“Which part of ‘all’ did you not understand?”)

Typos, OK. Synonyms, OK. But isn't there a point at which the user has supplied sufficient signals to convey the message This Really Is What I Meant And I'm Not Kidding?
</id>

According to Danny Sullivan the super synonyms / neural matching algorithm is being used on about 30% of searches.

I have to say that every day I search for stuff and get frustrated.

Example: I searched for [celebrity name] [my keyword (which is a tool/meme)] the other day because I'd heard that a celebrity mentioned using [my keyword] on a chat show a few weeks ago. The first page of results contained variations of the following: tools/memes matching [my keyword] with no mention of celebrity, [my keyword] articles about a different celebrity, an empty, unanswered forum post asking about [celebrity name] using [my keyword], and a long forum post about a different celebrity using [my keyword]. I had to go to the second page of results to actually get a couple of articles about the celebrity in question talking about [my keyword]. If that isn't synonyms getting wildly out of control then something else is going badly wrong.

Bing and DuckDuckGo both gave me the answer I was looking for immediately, at position one in the search.

I keep seeing this again and again. I keep having to go to Bing or DuckDuckGo to find what I'm actually searching for. Some of the oldest and most original tools and memes in my niche have been wiped completely from the results, despite having huge backlink profiles. I'm seeing some crazy injustices, where people who've created second rate copies of famous tools/memes are getting all the credit for them, while the originals are disappearing down the serps.

Funnily enough, the Google results had changed today to at least bring the articles I was looking for onto the first page of results, (although they are still not at the top) so I guess the algorithm could be learning from user experience.

^Maybe that's why my hobby site is down 30% organic traffic, LOL

I wonder if synonyms / super synonyms / neural matching sells more ad clicks...

I wonder if I bid on [widget] and someone searches for -synonym of widget- my ad for [widget] gets shown...

I'm sure that if I bid on -widget- and someone searches for -synonym of widget- my ad for -widget- gets shown...

^Maybe that's why my hobby site is down 30% organic traffic, LOL

I honestly think that's what's happened to a lot of people, particularly the long established businesses that are suffering this year. For some reason, perhaps for technical SEO reasons like keywords and social tags, or just because they weren't ranking in position 1 or 2, they're getting excluded or pushed down the serps in the super synonyms searches in favour of other sites that rank highly in their own right, but are wildly off-target on subject matter.

I've tried a few different searches, and the algorithm seems to return position 1 and position 2 for my short tail keyword for almost everything whether it's relevant to the specific search or not, then a bunch of stuff from left of field that has authority behind the site (empty pages included), then lower down and even on page 2 shows stuff that actually fits.

Thinking about it, it may even be why people are complaining about how they're losing out to big brands, and how big brands are ranking above them even when they have discontinued products or thin pages.

Neural ... or the fact that most folks surf and are personalized and g presupposes what they want?

Make the same queries in private mode (clear the cache first) and lock down all scripts NOT REQUIRED and see what kind of results you get.

There's two sides to the serps, and both of them are generally not in your control. Just sayin'

YMMV

I'm sure that if I bid on -widget- and someone searches for -synonym of widget- my ad for -widget- gets shown...

Yes, it's called "broad match" and while sometimes it's a good thing it can waste a lot of ad dollars if you're not careful. An advertiser using broad match is well advised to develop a long, strong list of terms to exclude.

As Aristotle pointed out above, the results degrade when things are matched that aren't really legitimate synonyms.

"Wedding gown" and "bridal gown" are good synonyms. "Evening gown" and "night gown" are not.

That's not a made-up example. It comes and goes but I just checked and one of the organic results for "night gowns" was Bloomingdale's main page for evening and formal gowns.

Yes, it's called "broad match"

In my (rhetorical) questions I addressed that by using the [exact match] brackets in one and -broad match- in the other.

Follow the money.

Night Gown: 40,500 Average Monthly Searches, $0.49 Top of Page Bid
Evening Gown: 74,000 Average Monthly Searches, $0.82 Top of Page bid

Why modify a searcher's query based on synonyms / super synonyms / neural matching?
Could be as simple as the old "people also bought" cross sell, or in this case simply pushing the searcher towards "higher value" searches by suggestion.
Could be more nefarious, a workaround against [exact match] and -negative keywords in the ads
Could be just another way to water down the organic results to make the ads more appealing.

Could also be that the average searcher doesn't know the difference, and possibly regional variance on what the items are commonly called.

I noticed the term "super synonyms" being used in this thread, and wasn't sure exactly what it means. So I did a google search for it, enclosing the whole term in quotation marks.

But despite the enclosing quotation marks, all of the results on the first two pages assumed that I was looking for synonyms of the word "super". In fact, I had to go to the last result on the third page to find an article about google's use of "super synonyms". Oddly it's a linkedin article:

[linkedin.com ]

It seems rather ironic to me that google's algorithm so badly mis-interpreted google's own invented term "super synonyms".

@Shepherd, if you look at the financials, isn't the answer (no pun intended) kinda obvious?

or in this case simply pushing the searcher towards "higher value" searches by suggestion

What's the point of that when it's not what the user was seeking?

Could also be that the average searcher doesn't know the difference, and possibly regional variance on what the items are commonly called.

In this case it was Google that didn't know the difference.

My dictionary's definition for synonym:

"a word or phrase that means exactly or nearly the same as another word or phrase in the same language,"

What's the point of that when it's not what the user was seeking?

Same reason the milk is always in the back of the grocery store.

In this case it was Google that didn't know the difference.

Maybe I'm misreading how the thread's going, in case you think I'm implying you don't know the meaning of words :)

I was just adding to Shepherd's list of reasons, and given that personalisation and user behaviour play a part, other people (and by extension the Google algo) could potentially distort the intended meaning of things.

I'm trying to understand what's going on. Do you think they would unleash a machine learning algorithm into the live search environment, that's trying to understand what people want when they type in an ambiguous query? If the algorithm didn't have enough information or limitations it would explain the weird traffic behaviour people are seeing, where their pages are popping in and out of the serps or they get abnormal traffic spikes or drops despite their typical keyword rankings not seeming to change?

Is that a possibility or too far fetched?

Do you think they would unleash a machine learning algorithm into the live search environment, that's trying to understand what people want when they type in an ambiguous query?

Of course. That would be a lot quicker (and more in tune with the real world) than testing behind a firewall.

buckworks wrote:
My dictionary's definition for synonym:

"a word or phrase that means exactly or nearly the same as another word or phrase in the same language,"

People say that the term "guinea pigs" is a misnomer because
1. They aren't pigs.
2. They don't come from Guinea.

Likewise, the term "super synonyms" is a misnomer because
1. They aren't synonyms.
2. They aren't "super".