MSN Search Technology Preview -- Round 2

[Sharky]

Check out the "second peek" at:

[techpreview.search.msn.com...]

I heard that they have more than 5B pages in the index this time.

[Imaster]

Its on and off. Now I get The MSN Search Technology Preview is currently unavailable. The site may be down for a short while due to scheduled maintenance. Please try again later.

[steveb]

The thing is back online.

I didn't do a lot of site: type searches a few days ago, but for two sites it is returning 1/10th the pages as a couple days ago.

[Spica]

I really like the results for common searches. Unfortunately, the search engine is currently worthless if you are looking for scientific knowledge. In specialized areas, MSN has a long, long way to go if they want to catch up on Google.

[Receptional]

The site seems online for .com but not up for other countries right now. Oops - no - now it is down... Yoyo here :)

In specialized areas, MSN has a long, long way to go if they want to catch up on Google.

My guess is that MSN are working hard at the high demand / popular search term end to ensure that spammy results do not get through. This might indeed be dampening niche phrases, but that may also be a function of MSNBot just not being around long enough yet to understand the good niche sites.

I guess if you are seeing that Spica, it may also suggest that MSN are using a theming algorithms - like Latent Semantic Indexing and Inverse Document Frequency [webmasterworld.com] which may not pick up the relationship between, say, Benzene Rings and Ferrous Oxide (or weirder relationships) and thus cannot theme quality sites of a reference nature?

Just me thinking allowed - I haven't tested the theory or anything.

What I did notice in a specialized search (although it was not up long enough for me to fully test) was that large sites with flat URLs fair well - didn't see anything with a query in the URL, but the first seven or eight were all clearly database driven sites.

[edited by: engine at 10:44 am (utc) on Oct. 7, 2004]
[edit reason] fixed url [/edit]

[chrisnrae]

Yea, I got in tested like 5 searches and then it cut me off ;). I am still seeing a lot of www.domain.com/?affid=idhere taking high results on competitive SERPs. IT isn't taking 4 or 5 of the top 10 like it was before, but it still looks a little cheesy to me. Also seems very "transparent". Personally, my own things are good to go, but the index as a whole needs some work. Much better second round though - but I certainly hope there is a third expected if they want to give people a reason to switch. ;)

[Livenomadic]

My Main Keyword:

Google: 983,000 results

MS Search: 28,952,938 results

Looks like we have a war on our hands... :)

[Macro]

In several search terms I'm ranking #1 out of 20-30 million pages.

This has gotta be a good SE ;)

[sean]

msndude, I might have missed something. Is your team working on any breakthrough features that would give people a clear reason to to switch to MSN Search, i.e. Google's Pagerank in the late 90's, or for now is it a matter of tweaking the existing 'knobs' to achieve a better balance?

[george123]

"In several search terms I'm ranking #1 out of 20-30 million pages"
me too for the KW :"hsgiagf cuiq8wt`ygugdiugd&&&hagduyqg"
at the moment is just a c.......p ...those zillions SERPS are not THERE.....
I made a search of a very distinguised country and the result was 20-30 million pages (:DE) ,by the way,that country does not exist....today...,do they crowl the EB or the GBofR from the begining of the century?

[bhartzer]

It's interesting to note that I'm finding sites that I know are banned in the current Inktomi/MSN results are not banned in the tech preview.

Perhaps they haven't applied any/all spam filters to these beta results?

[airpal]

this-domain-will-rank-higher-than-all-other-sites-because-of-the-domain-keywords-and-paid-for-or-blog-spammed-thousands-of-links-muhahahah (dot com). That's all I have to say. Awaiting techpreview 3.

[bhartzer]

airpal, LOL!

You're absolutely correct!

[skunker]

msndude,
Again, thanks for letting us monitor the new search engine.

I like the results, however, I do have one complaint. It seems that the search results are not THAT accurate. For example, lets say that I am creating a site about The Alamo in San Antonio, Texas. Well, when I do a search about that, I get Alamo car rentals, Alamo Draft Cinema, and PC Alamo, a computer company. The only real result about the alamo is like 5 results down.

This is not very accurate.

Other than that, I am happy with it.

[airpal]

"In November we will begin to migrate to our own algorithmic search results."

Uggh. They better address the domain keyword weighting at the least! Also, there seems to be no way to distinguish one link as more valuable than another. 500 Incoming sitewide links from an external domain seem to weigh as much as 500 incoming links from different domains.

Then they can try and lure a Google Engineer to learn how to block the subdomain spamming and extensive crosslinking on the same IP address.

After that, they will be ready to take on Google.
Even Yahoo properly implements incoming anchor text links from different domains as the #1 ranking factor, but haven't addressed dupe content or subdomaining/crosslinking as effectively.

[cleanup]

I think the SERPS look excellent. The results look clean and understandable not like the hotch potch mixture and lack of returns that Y! are providing.

Just one point.
There seems to be a duplicate issue. I have a .net and .com domain both pointing to the same place and results were displayed for both, Google took a while to get on top of that as I recall I hope MSN can sort it out with less fuss/collateral damage.

[airpal]

[I think the SERPS look excellent. The results look clean and understandable not like the hotch potch mixture and lack of returns that Y! are providing.]

Let me guess... your pages must rank horribly in MSN, but their algorithm is still great, right?

(you need to relax... I'm working on an enormous new site so I've seen no results on Goosandbox, Yahoo or MSN yet. either way, it's nice to be ranked #1 on an SE, but that does not make that SE a relevant search engine)

[edited by: airpal at 4:24 pm (utc) on Oct. 5, 2004]

[cleanup]

Yeah great post! airpal.

And yours bomb on Yahoo! I asume.

I like the results
1)because they return Google-like results (and I always use Google)

and

2)because they return more results, so for obscure searches where say Google only returns 2 pages MSN is returning 3 or 4 pages of SERPS.

I have actually gone to the new MSN twice today to look for "people names" I was researching and found one that I could not find on Google. Needless to say I do not bother with Y!

[edited by: WebGuerrilla at 5:52 pm (utc) on Oct. 5, 2004]
[edit reason] keep it polite please [/edit]

[Imaster]

301 redirects don't seem to be working.

[papamaku]

The estimate of the number of results is very dodgy:

Search for microsoft: 89,731,468 results
Search for microsoft car: 89,377,531 results
Search for microsoft porn: 89,377,531 results
Search for microsoft linux: 89,377,531 results
Search for microsoft -car: 89,729,023 results

[Imaster]

Its broken I guess, check this out

Search for microsoft : 89,731,468 results
Search for microsoft -msn : 89,733,512 results

Now how could the second query have more results than the first one?

[The Grizzler]

Interesting...

A search for the word "the" or "http" on techpreview comes up with 2,000,000,000 pages. This is obviously a capped figure. I would think that at the rate msnbot appears to be crawling, it will have indexed a much greater number of pages than Google by the time it goes live (and probably already has).

I spent the majority of last night putting msn search preview #2 through it's paces and it is far more accurate than Google for both specific and general searches, particularly searching by location.

IMHO results are so much better it will only take searchers a couple of visits to the new msn search to think about making a switch.

Google beware.

[olias]

301 redirects don't seem to be working.

Really? I was going to point out that msn seem to have handled my 301's perfectly well unlike both Yahoo and Google.

[webmktg]

Rankings are tottaly different between the beta version and the current msn.

[webmktg]

Rankings in new msn beta is not good for me. Although I have many content rich websites they are not showing up for searches.

Anyway my sites are listed with new msn beta version. Does this mean that I still have a chance of getting top rankings?

[webmktg]

I think msn beta has many pages. For one one of my main serach terms I got the following

msn beta 61,586,414
msn 950,111
Google 5,770,000
Yahoo 4,890,000

New msn beta version is searching more documents, which means increase competition.

[webmktg]

Another finding

I have noticed that a list of websites in my industry targeting different destinations are coming up on top in msn beta version. These search terms are highly competitive in my industry.

The funny thing is these sites have no Page Rank (PR is Zero) in Google. Also they are not content rich websites.

I have following questions for you guys

What does this mean?
Why these sites ranking wel in msn beta version?
What would be the ranking criteria in msn?

[Spica]

In a search for an obscure scientific word, the new MSN claims about 1,576 results, but let me see only 20. I hope the limitation is only in the beta version. Otherwise, what's that all about?

[pleeker]

I'm with steveb and Chndru from several days ago ... still a long ways to go. How do you explain that eBay isn't in the top 100 on a search for ... that 7-letter word which describes what happens at eBay? And is only #11 on the plural version of that word? (Trying to tip-toe across specifics....)

Checked one of our clients main keywords and the #1 site listed is still the same one I mentioned from preview #1 -- a page made up of one huge image and no text other than a copyright notice at the bottom. The only thing possibly justifying the page's top position is that it was created with Front Page. (That's how you optimize for MSN Search -- build it with Front Page, submit it from a Hotmail address, etc.) :)

Admittedly, some test searches I've done yesterday and today are an improvement over the summer, but this is far from being ready for launch.

[mfishy]

A couple general examples of how horrible this engine is.

Books - No Amazon
Search Engine - no Google, Yahoo, MSN

[Receptional]

Books - No Amazon

Suits me! :)