From Mexico.purplecow.org
(tmpsave b4 lunch) |
(finish draft) |
||
| Line 1: | Line 1: | ||
| - | |||
= IN PROGRESS = | = IN PROGRESS = | ||
| Line 18: | Line 17: | ||
I'm figuring the average user doesn't look at more than three, but six sounds about fair. "Top half of the page" and all, although probably not on a 600px tall nyetbook screen. | I'm figuring the average user doesn't look at more than three, but six sounds about fair. "Top half of the page" and all, although probably not on a 600px tall nyetbook screen. | ||
| - | == Test 1 == | + | === Test 1 === |
| - | Question: "Can I take a camera to the Meredith Music Festival?" | + | * Question: "Can I take a camera to the Meredith Music Festival?" |
| - | Search Terms: <tt>camera rules meredith music festival</tt> | + | ** Search Terms: <tt>camera rules meredith music festival</tt> |
| - | Bing: Exceed. | + | ** Bing: Exceed. |
| - | Google: Exceed. | + | ** Google: Exceed. |
| + | |||
| + | === Test 2 === | ||
| + | * Question: "I need specs on the 32GB Transcend ExpressCard" | ||
| + | ** Search Terms: <tt>transcend 32gb expresscard</tt> | ||
| + | ** Bing: Pass, rank #3 | ||
| + | ** Google: Pass, rank #2 | ||
| + | |||
| + | === Test 3 === | ||
| + | * Question: "What is a Sun part # 501-4489?" | ||
| + | ** Search Terms: <tt>sun 501-4489</tt> | ||
| + | ** Bing: Exceed. | ||
| + | ** Google: Exceed. | ||
| + | |||
| + | === Test 4 === | ||
| + | * Question: "Where's that Verge restaurant someone was telling me about?" | ||
| + | ** Search Terms: <tt>Melbourne Verge</tt> | ||
| + | ** Bing: Pass, although with a sponsored link for another restaurant under a Verge header. Not good. | ||
| + | ** Google: Exceed. | ||
| + | |||
| + | === Test 5 === | ||
| + | * Question: "Where are the docs for upgrading Sun VDI Broker?" | ||
| + | ** Search Terms: <tt>Sun VDI Upgrade</tt> | ||
| + | ** Bing: Fail. | ||
| + | ** Google: Pass. | ||
| + | |||
| + | === Test 6 === | ||
| + | * Question: "I want to find a starting point for the AlienFX API." | ||
| + | ** Search Terms: <tt>AlienFX API</tt> | ||
| + | ** Bing: Pass. | ||
| + | ** Google: Fail. (with malware sites ranking) | ||
| - | == | + | == Conclusion == |
| - | + | ||
| - | + | ||
| - | + | ||
| - | + | ||
| - | + | They're roughly the same. Opera's web debugging tool indicated that time to run the search was within the margin of error for the two - so no performance winner there. | |
| - | + | ||
| - | + | ||
| - | + | ||
| - | + | ||
| - | + | For obvious searches, the top results are the same. The VDI search failed for Bing because it turned up forum posts ahead of actual content and Google returned Malware for the AlienFX search. In general, I think being able to weed out malware is probably a good thing. | |
| - | + | ||
| - | + | ||
| - | + | ||
| - | + | ||
| - | + | [[Category: Tech]] | |
Revision as of 02:31, 5 October 2010
IN PROGRESS
Bing vs Google
Overview
Most formal tests seem to focus on highly specific (crafted?) searches designed to favour one engine or another. Not very scientific.
So, to make it _less_ scientific, I have decided to compare results for some searches that I actually ran today.
Results Key
EXCEED: Correct answer at #1 position PASS: Correct answer in top six results FAIL: No useful answer in top six results
I'm figuring the average user doesn't look at more than three, but six sounds about fair. "Top half of the page" and all, although probably not on a 600px tall nyetbook screen.
Test 1
- Question: "Can I take a camera to the Meredith Music Festival?"
- Search Terms: camera rules meredith music festival
- Bing: Exceed.
- Google: Exceed.
Test 2
- Question: "I need specs on the 32GB Transcend ExpressCard"
- Search Terms: transcend 32gb expresscard
- Bing: Pass, rank #3
- Google: Pass, rank #2
Test 3
- Question: "What is a Sun part # 501-4489?"
- Search Terms: sun 501-4489
- Bing: Exceed.
- Google: Exceed.
Test 4
- Question: "Where's that Verge restaurant someone was telling me about?"
- Search Terms: Melbourne Verge
- Bing: Pass, although with a sponsored link for another restaurant under a Verge header. Not good.
- Google: Exceed.
Test 5
- Question: "Where are the docs for upgrading Sun VDI Broker?"
- Search Terms: Sun VDI Upgrade
- Bing: Fail.
- Google: Pass.
Test 6
- Question: "I want to find a starting point for the AlienFX API."
- Search Terms: AlienFX API
- Bing: Pass.
- Google: Fail. (with malware sites ranking)
Conclusion
They're roughly the same. Opera's web debugging tool indicated that time to run the search was within the margin of error for the two - so no performance winner there.
For obvious searches, the top results are the same. The VDI search failed for Bing because it turned up forum posts ahead of actual content and Google returned Malware for the AlienFX search. In general, I think being able to weed out malware is probably a good thing.
![[]](/skins/blender/open.png)
