Yahoo Speculation, INK and AllTheWeb and AV, Oh My! Options

[Grumpus]

I sort of got us going in an off-topic direction in this post about Meta Tags. Since this is a hot topic right now (and will likely grow even more over the coming weeks/months) I thought I'd split it off so we can discuss Yahoo's future here without totally waylaying the Meta Tag Discussion in the other thread.

To start us off, here's what Qwerty said:

I believe they bought Overture more for PPC than to get ahold of AV and ATW (which Overture owned), but they've got them, and it makes sense for them to use them. Maybe not right away, but I think eventually we'll see ATW's index ranked by INK's algo (so hopefully a lot of the trash in there will go away), and maybe AV's image search, which I find much better than Google's.

I've been saying for a long time (July of 2002 is the first time I said it - over in WMW) that all signs show that Yahoo really wants to be out of the search business altogether. The problem is that they really haven't been able to afford to be out of it - that is, until now.

October's purchase of Overture was the key to it all. The purchase of INK was important, but just having crawled search capabilities does not a search provider make. (Even Google uses the DMOZ as part of their search service).

You'll notice that with all of the purchases and partnerships Yahoo! has made over the past half decade or so, they instantly slap their brand on it (like eGroups/groups.yahoo.com or the partnership that gave us Yahoo!/SBC High Speed Internet). The exceptions to this rule have been when the purchases are in the search sector. Visit an INK site and you'll be hard pressed to find the word Yahoo! anywhere. Visit the Overture site and you can see it on the corporate page, but that is it. They simply are not looking to put their name on search services - it would, afterall, dillute their brand by having Yahoo! plastered all over the place on MSN (using INK), AllTheWeb, AltaVista, all the sites using Overture, and so on.

It's far better that they use the "Overture" brand (or one of the other brands) for search, and the Yahoo brand for the plethora of things you will find on the Yahoo site. (the same way that ABC is owned by Disney, but maintains its own separate entity.) It makes sense, then, that Yahoo would provide a whole line of services on their site with "Search Services provided by Overture."

Yahoo, back when INK was first purchased, never really said exactly what they were going to do with it, but insinuations were made that it would be developed into a corporate search solution - and they most assuredly never said that it would be used for serving their own Yahoo! Search results.

That's not to say that the INK algo won't be incorporated, in part, into the results generated by AlltheWeb or AV. (And if you look at the AV results right now, there are some signs that they may already be experimenting with this).

To be truly prepared for what's to come from Yahoo, I do think it would be a mistake to ignore Inktomi completely, but at the end of the day it is what we see coming off AV or AllTheWeb that we'll need to comprehend to rank well at Yahoo. (I believe that AllTheWeb is the "experimental" engine and that AV is the "public consumption" version - though I may have that backwards... )

Yahoo might also use INK to provide a "middle level" of results in their new plans. (i.e. When you search you get "Directory Sites" "Overture PPC" "INK PFI" and then the "free inclusion backfill" coming off ATW/AV. I have a feeling that wouldn't go over very well as there would be just too much fluff up top before you got to the generic results.

So, my conclusion (though it can be, by no means, considered fact) is that we're going to have to really have our brains wrapped around the notion of ranking well with AllTheWeb/AV - and I'll see if I can find which is the experimental one - one of them will give us foreshadowing of "things to come" and the other will give us a look at "what is"). If any of the INK technologies are used in the generic results, that will show up on the experimental engine first and INK will never be used (as an entity) to provide Yahoo's generic search results.

Maybe.

G.

[Grumpus]

Also in that original thread that I inadvertantly hijacked, Jill said:

So I think it's safe to say that Grumpus's 2004 prediction is that Yahoo will eventually begin to use ATW and/or AV technology in their search results.

Actually, that's not my 2004 prediction. It's a prediction from February 19th of 2003. In a discussion on that date over at cre8asite forums I said:

Here's what is likely going to happen over at Yahoo during the coming months. Yahoo will stay with Google through summer. Yahoo will keep using the Overture results and we'll see a steady upward trend as the two get more comfortable in the same bed. Meanwhile, Overture will be out there with a chainsaw and sledgehammer getting AltaVista so that it actually produces a valid and viable list of web pages. (This may be a little bit out on a limb, but here it is anway: Look for a new and shiny "Pay For A Deep Crawl" concept sometime this spring or early summer).

Once Overture has a viable directory and page search thing going, Yahoo will fluff the pillows and hand over the entire "web search" aspect of their business over to the Overture guys. (Somehow, Overture will end up as Yahoo's directory - whether Yahoo sells the directory to them or whether it just rather goes away and the people who paid for getting in there will get listed in Overture for the duration of their contract? I dunno. Something along those lines...)

Regardless, Yahoo has definitely shown signs of wanting OUT of the web search biz for some time now. This deal gives them a viable means of doing so.

The timetable is way off and I shouldn't have done the "go out on the limb" part, but everything else is fairly uncanny, no?

G.

[Jill]

FWIW, I spoke to Yahoo reps on the phone soon after the Overture purchase was announced. What I learned was summed up in this little blurb in the Advisor, entitled, Yahoo to Purchase Overture.

The gist was that they fully intended to integrate all of their new search properties into Yahoo Search, but only when they felt the relevancy was as good or better than what they had with Google.

With Google's not so relevant results at the moment, this would certainly be a good time to get with the program. If they don't do it soon, then they must be still having relevancy problems of their own.

Jill

[rustybrick]

Jill and others,

I am sure you remember the talks at SES Chicago about why Yahoo hasnt made the big switch yet.

(1) Complication of Integration
(2) Legal Issues with Google
(3) Politics (which engine technology to use)

So far we have seen hints of Inktomi in the results, even more so overseas.

[Grumpus]

I think the timing of Yahoo's announcement will be critical. Google's IPO is going to cause a lot of noise. If they go too early, then Google's announcement will come in and steal the attention needed to make it all go well. If they time it right, though, they can hit right at about the same time and, in effect, steal some of Google's thunder rather than the other way around. (And Google's thunder, right now, is just a tad louder than Yahoo's.)

G.

[Grumpus]

So far we have seen hints of Inktomi in the results, even more so overseas.

There are hints of INKtomi in AV's results, too.

Also, I was on a meta engine the other night (I'm bummed, because I can't, for the life of me remember which one) and it offered the ability to tab between most the major search engines. On this night, if my search term was specific (one obscure word or multiple words) the Yahoo and AV tabs were identical. If I used a very broad term (I think one of the ones I used was "Jobs") then they didn't match (nor did they match Google's results).

Interestingly, when I went to Yahoo and AV respectively and ran the same searches that were matching on the Meta, neither were even CLOSE to the results being produced by the meta engine. This leads me to believe that the package that Yahoo is marketing to meta engines (and others who want their search results) is already incorporating the Overture run technologies into the search.

G.

[qwerty]

But nobody outside the Googleplex knows when (or even if) the announcement will be made. It's still very much within the realm of possibility that Google will not make a public offering this year.

Yahoo should act soon, and make a big deal about it. IMO, they should have done it about a month ago. My guess is that they're still testing things. They want to be able to confidently say that the "new and improved" Yahoo search is significantly more useful than that mess Google had been providing them with.

The INK/AV/ATW mix prediction I'm quoted as making is not something I expect to see right away. I think it's more likely that Yahoo will come out with a new search using INK results in the short term and will over time bring in some of their other technology. The people who've been saying they sometimes see INK results in Yahoo have not, AFAIK, indicated there was anything other than INK in there.

[Grumpus]

I found the meta site I was using last night. It's good because you can easily compare rankings side by side. And it looks like you may be right - sort of.

The meta site is at websearch.com <- A keyword domain name that I forgot. (IMG:http://www.highrankings.com/forum/style_emoticons/default/wink.gif)

Let's plug in "highrankings" as the search term.

If you switch between Yahoo and Altavista - the results are identical. Interestingly, those results are also identical to the pure INK search at positiontech. So, AltaVista is (in this case, anyway) feeding INK results to the meta engines. (The same search on altavista, doesn't produce the same results, nor does it on Yahoo).

So, when a meta engine purchases a feed from Yahoo and AV, they are actually getting an INK feed? That looks to be the case - at least here.

Curiouser and curiouser.

G.

[rustybrick]

Interesting stuff....looking like Inktomi.

Any evidence of AV or ATW results anywhere?

[Grumpus]

On that websearch.com site, you can get different results between the AV and Yahoo tabs by using a term that you might expect to find more "spammy" results (i.e. More comercially competitive - I just tried "Hotels" and that works...).

Haven't had a chance to asses what's bringing those results out, though. Will check in a bit when I get some time. (Maybe those are coming from Overture in some way? I have no idea...)

More in a bit - unless someone else finds the answer first...

G.

[Ron Carnell]

Yahoo! has never confirmed how much stock they own in Google, but estimates run between five and ten percent. With that kind of investment at risk, they won't do anything to hurt Google's IPO unless it give them a substantial and immediate edge.

Would you? (IMG:http://www.highrankings.com/forum/style_emoticons/default/smile.gif)

[Grumpus]

If it would cut my own direct losses (or improve my gains), you bet I would. I though, like you, don't know how much that interest in Google really is. At 5%, you can bet I'd be riding shirttails rather than trying to go out on my own too early. At 10% or more, it might start being less of a difference in my bottom line no matter which way I went...

G.

[edge]

I might be wrong but in looking and comparing results with Google and Yahoo, they are very different as of now. On a few forums as of late Ive seen Ink show up as the supplier for Yahoo.

Anyone else seen this?

[Grumpus]

There are three sets of results appearing in Yahoo. Which set you get varies from keyphrase to keyphrase, hour to hour, and day to day. One is still Google (which you'll get fairly consistently by using oddball terms ("highrankings" is an example). The second set is definitely INK as it matches up exactly with the PositionTech Pure Search tool.

The third set, though, is a mystery. It's not either of the above, it's not AV or AllTheWeb, nor is it even Overture. It may be an infusion of some or all of these (including Google, for that matter) but it's definitely not something we can observe with any consistency to even begin to guess how it might rank things.

In fact, I can't even be certain there are only three sets - that third set is so unidentifiable, that there may actually be several of those showing up and I just don't know it.

G.

[edge]

I was starting to wonder how many sets and how often they changed. I went to show someone a result set only to have it 80% different :) Kind of hard to prove a point when it changes. With lack of knowing about ATW and AV, I wouldnt of thought to check them. Thanks Grumpus


At least all my work to get better results in INK will pay off now.