|I would much rather prefer Google not to try play 'quality editor' |
Google's business is predicated on them being exactly that, a quality editor. Whether we like it or not Google is in the business of giving their opinion of what constitutes relevance.
In a recent court case revolving around their ranking system (Google was the defendant), Google argued (if my memory serves me right) their case from the angle of freedom of speech, and that PageRank was part of an expression of their opinion.
|I think it's a highly irrational system that loves to pretend it presents quality sites but I think that's rubbish. |
"I'm not saying I disagree with you- But I'm not saying I agree with you, either"
That's a quote from Jim Thompson's wonderful novel, "pop.1280" that expresses how I feel about this post.
[edited by: martinibuster at 4:05 am (utc) on May 22, 2003]
It really doesn't matter whether the Spice Girls make good music or not, what does matter is that a search engine bring up relevant results when someone is looking for Spice Girls. If searchers think the music is good they will buy and if they don't, they won't buy. Search engines are not responsible for sites' conversion to sales, though I understand that Google Adwords is very helpful in that respect.
Of course, if a search engine is not delivering relevant results we can always choose to use another - that's the beauty of the internet; we can surf to or surf away any time we please.
If Google's algorithm and system of ranking are so inadequate, I assume you've thought of what you consider a better way to rank sites. We like to hear constructive suggestions and particularly enjoy learning about new algorithms and search engine research.
Tell us your ideas of how it should be done.
|I would much rather prefer Google not to try play 'quality editor' which is, in my opinion, just a marketing scheme since the requirements for such are impossible yet. |
Aesthetics isn't and shouldn't be the domain of search engines, who have no means of measuring such. That's the reason those irrelevant criteria are there in the first place.
I didn't know google was trying to make aesthetics part of its quality control. You'll have to clarify what you mean by that. Google only cares about the content and tries (with some success) to deliver the right content pages someone is looking for. The incoming links criteria has worked, because it usually suggest that those are pages that are popular, therefore they are more likely to provide quality answers than those that are never linked to.
|Quality selection requires a very advanced sort of A.I. which is currently not available. |
Precisely. That's why what we have now will have to do. :-)
|If Google's algorithm and system of ranking are so inadequate, I assume you've thought of what you consider a better way to rank sites. We like to hear constructive suggestions and particularly enjoy learning about new algorithms and search engine research. |
Tell us your ideas of how it should be done
By a Human. A bot cant see, or think for that matter. A human can say this site is relevant, a bot can scan for the rest. A company can spend $1,000,000 on a site, and its really awesome, it has info up the kazoo. Extremly relevant and easy function, and helping tons of folks.
But no one will find it without tricks, and links........
"There are 3 billion sites"! Some one will say. Well lets get started looks like we have our work cut out for us.........
"It really doesn't matter whether the Spice Girls make good music or not, what does matter is that a search engine bring up relevant results when someone is looking for Spice Girls."
This is a faux pas, since 'relevant' already is a quality (and therefore an aesthetic) criterium. Google currently shows what is popular, not what is 'relevant' - relevant to whom, anyhow?
Do you want to try argue with me about the top sites that show up when I do a search on <snip> - the first result is the biggest scam corperation which ever tried to pollute this art.
To whom should that result be relevant? Relevant in what way?
[edited by: Marcia at 4:41 am (utc) on May 22, 2003]
[edit reason] No specifics, please. [/edit]
|By a Human. A bot cant see, or think for that matter. |
Added:(in two cat's bless their hearts)
yep! : )
"The incoming links criteria has worked, because it usually suggest that those are pages that are popular, therefore they are more likely to provide quality answers than those that are never linked to"
We agree to disagree on that one. The popularity of a thing says nothing whatsoever about it's quality. You can take my example, if you'd wish, and explain why a company who scams untalented people out of their money deserves a top ranking on the word <snip>....be my guest.
[edited by: Marcia at 4:43 am (utc) on May 22, 2003]
[edit reason] No specifics necessary. [/edit]
You have three choices. You can either dance to the tune of the fiddler. You can walk out of the pub in disgust. Or you can slip the guy a "Mickey Finn".
The incoming links worked until google chose to put pagerank on the toolbar. If webmasters did not know the PR of a site they would not be able to use that as a tool to judge the relevancy of links and would have to judge on perceived quality.
In my opinion - putting PR on the toolbar was a huge mistake that ultimately can only further & further reduce the effectiveness of their algorithms.
What option do you suggest for thaose of us that want to find relevant sites on the web?
As you already implied that what you want is impossible, what do you suggest?
|who really believes the Spice Girls make great music because a lot of people buy their records? |
Probably the millions of people who bought their CD.
Tastes vary. What's good to you may not be to me and vice versa. The only way of tracking a general trend is popularity. "If this many people like it, it must be a good place to start." - that sort of mentality.
I think it works quite well. At least until google finds a way to determine which sites I like best and compare their database to that and give me personalized results.
I'm not disgusted with the current state of affairs. I am, however, interested in a discussion about what criteria could be thought of that'd make more sense than just H1 or domain names or incoming links.
I'm not sure I'd trust a human editor any better than an algorhytm either. In fact, I'd much prefer the latter for web indexing as my faith in humanity is rather low and I think DMOZ is the perfect example of the 'Makable Society' of social democrats.
I'm afraid Google has become just another bureaucrat's dream. You have a bunch of criteria, you pass, you're the model citizen we've been looking for. We want better citizens. So we need more criteria. And so on and on.
You sound like someone who has a low pagerank :-)
I have MANY ligitimate complaints against Google, one of which I thought you were going to address when I first saw the subject title of this thread.
PR is not one of my complaints. I fully understand what you're complaining about, but I don't toally agree.
How do you suggest Google should rank sites?
Without page rank there would be too many crappy sites stuffed with keywords ruling the serps(Even with page-rank we still have this problem). The idea behind page-rank is that good sites naturally will have more people linking to them.
I agree PR has its problems, but I honestly don't see a better alternative.
One interesting tidbit is that PageRank!= popularity. It's closer to importance. For example, folks may visit lots of porn sites, but typical web sites don't link there. It's a little like measuring what TV shows people watch versus what they report they watch. The latter skews more toward public broadcasting and the former skews more toward Jerry Springer. :) For web search, you probably want the public broadcasting more than the "My cousin slept with my wife, then ran off to join the circus!"
I wonder if this post will make as much sense when I read it tomorrow. Probably not. Oh well.. :)
What I want is simple: stop all that bull**** talk about 'quality content' and 'relevant content' and call the bird by it's real name: popular content.
>>In my opinion - putting PR on the toolbar was a huge mistake that ultimately can only further & further reduce the effectiveness of their algorithms.
[Edit - I love the PR on the toolbar. However it does seem to encourage "PR based link trading", etc..)
[edited by: John_Creed at 4:46 am (utc) on May 22, 2003]
|Quality selection requires a very advanced sort of A.I. which is currently not available |
oraqref you are right.
They are working on it. They are still a young company ;)
Yet you would not want to compare the search quality of Google with that of Altavista a couple of years ago.
Its the best thing aroung to continously get 250 million searches a day [webmasterworld.com]
>>I am, however, interested in a discussion about what criteria could be thought of that'd make more sense than just H1 or domain names or incoming links.
One of the problems is that a link is seen as a vote, but is difficult to recognise as a "negative vote" - if someone criticises your webpage and references is it with a link.
Another is that if I have nine votes on my page towards quality resources and I add a tenth vote, that does not mean I value the previous nine votes less by adding that extra vote.
[edited by: vitaplease at 4:52 am (utc) on May 22, 2003]
"One interesting tidbit is that PageRank!= popularity. It's closer to importance. For example, folks may visit lots of porn sites, but typical web sites don't link there."
You've not been visiting a lot of weblogs, that's for sure!
"It's a little like measuring what TV shows people watch versus what they report they watch."
Did you ever actually bother to do a search on Jerry Springer? Can you explain what relevance the second result has?
Geez, oragref, have a little respect. He's not going to answer if you just yell at him.
Added: Well maybe he'll answer but show some respect anyway. He didn't personally trash your site.
"Geez, oragref, have a little respect. He's not going to answer if you just yell at him.
Added: Well maybe he'll answer but show some respect anyway. He didn't personally trash your site."
I don't understand your comments. I am not aware of any 'disrespect' in my reply to GG, and I'm not here because I either have low pagerank or someone trashed my site or whatever such. I asked him a simple question because we we're discussing relevancy and I thought it to be rather interesting that the example he used returned such strange results.
|What I want is simple: stop all that bull**** talk about 'quality content' and 'relevant content' and call the bird by it's real name: popular content. |
PageRank is only one of the 100+ factors that Google's algorithm weighs before determining where a page should be listed on the SERP. It isn't even the most important factor--it's more of a tiebreaker when everything else is equal.
Now let's deal with some of the other topics that were brought up in this thread:
DMOZ. You're comparing apples with oranges. DMOZ lists sites; Google lists pages. That's a big difference. Most people aren't looking for sites; they're looking for specific information, and that's where spidered search engines like Google come in.
Human-edited directories vs. spidered search. Even if directories tried to index every page on the Web, there's no way they could do so. If I publish an article on Napflion, Greece today, I can expect to see it listed in Google within 24 to 48 hours. There's no way any human-edited directory can respond that quickly--even if it's willing to include "deep links" to individual articles.
Human-edited directories and indexes certainly have their place, as do collections of links on editorial "content sites." But for sheer inclusiveness and comprehensiveness, they can't compete with a spidered search engine like Google.
I don't ever refer to them as quality content or popular content. I always call it search results.
If I can find something in the first couple of pages that is relevant to what I am looking for, I am generally satisfied. The more exact the search term the more relevance I expect.
The "popularity" or "importance" is in the context of your search. With a single word search it really isn't much of a help.
Suppose I want to find the volume of glacial ice in the Greenland icecap. A search on "ice" is not going to do me much good at all and the popularity will work against me on this one. But if I use search terms that will narrow the selection sufficiently, let's try "greenland icecap" and suddenly the more popular terms actually have some useful information.
And there really are a lot of variables that google looks at other than PR, H1, title, etc. They don't need hundreds of programmers just to do that.
Didn't mean to be unclear, oraqref. I mean that people's link patterns (what they decide to cite or reference) can be different than their surfing pattern (where they visit). So even if (for example) a porn site is popular, it doesn't mean that the porn site will have much PageRank.
<added> For what it's worth, I agree that search engines should have more sophisticated scoring using lessons from A.I. and machine learning</added>
I was referring to the
|stop all that bull**** talk |
It might not have been the best way to phrase what you wanted to say imho.
>One interesting tidbit is that PageRank!= popularity. It's closer to importance. For example, folks may visit lots of porn sites, but typical web sites don't link there.
Not exactly. The original academic paper refers to this being more as a representation of the likelihood that a random surfer would end up on a page. People also link to pages because they are popular, rather than just because the represent an authority. Why else is madonna.com a PR8, the same as ama-assn.org, the home page of the American Medical Association? It ain't that people think Madonna is an authority on much of anything, but that she is a well known pop culture star.
Google is #1 because they're the best, quite frankly, and a large part of that is due to the concept of pagerank. They're clearly not going to abandon that anytime soon.
But the beauty of the web is that there are scads of competing search engines and directories. You're free to use any of those instead of G, and if any come up with a truly great innovation then there's an excellent chance they will be a huge success.
Yup, rfgdxm1, I was just commenting on an interesting second-order effect. :)
OK GG, that I can agree with. A second order effect is that PR is correlated with importance. One other point here. One of the reasons that porn sites tend to have low PR is because there are so many different ones, and tend not to have "authority" status on the topic. Unlike say Harvard University does. People do link to porn sites a lot. It is just there are so many of them PR is spread over many sites.
| This 64 message thread spans 3 pages: 64 (  2 3 ) > > |