[Mycroft] [Statistics] January 2004, Week 2 (full week)

Ricky ricky at linuxbourg.ch
Mon Jan 12 11:35:04 EST 2004


Hello all,

Viva vivisimo! Vivisimo is the word of the week: new challenger, closest 
challenger, most error pages... you'll see vivisimo everywhere in these 
stats. Last week on rank 17 with 35 hits (5 per day) it ranked this week 
on rank 11 with ... 291 searches (over 40 per day), which is an increase 
of 830%!!! A closer analysis of the vivisimo phenomenon yields that most 
of the hits were made on Monday and Tuesday (see attached image), 
declining dramatically at the end of the week. This points to a very 
punctual action somewhere on the net. Unfortunately, I haven't been able 
to find anything interesting in the referrers.

By the way, the graph shows the two highest engines (dictionary and 
google), the current number 10 (dogpile, which has only a little bit 
more than vivisimo) and shows how much of their weekly hits they made in 
one day (expressed in percent). While the top 2 are very regular (around 
14% daily), dogpile is a little bit irregular (Monday the quicklink was 
launched a little bit late because of our CVS problem), while 
Saturday-Sunday shows some irregularities which I can't explain, but 
nevertheless, quite regular compared to vivisimo.

I've also checked the unique IPs and noticed that there's not just one 
person out there sending in hundreds of search requests in order to 
manipulate these stats. In fact, the ration unique-IP/Total-searches is 
quite the same for all 4 compared engines.

Since I've never used vivisimo, I thought I could have a look. To my 
surprize, on the frontpage you can read: "Mozilla Users - Install Search 
Plugin!" Yeah, that's right, they just copied Alan Bramley's plugin and 
modified it slightly. Here's the diff:

  <input name="query" user>
-<input name="v:sources" value="MSN,Netscape,Lycos,Looksmart,FindWhat">
+<input name="v:sources" value="Web">
  <input name="sourceid" value="Mozilla-search">
-
-#Interpret section doesn't work with frames
-#<interpret
-#    browserResultType="result"
-#    resultListStart="results retrieved "
-#    resultListEnd="Result Pages:"
-#    resultItemStart="  </span>"
-#    resultItemEnd="URL: "
-#>
  </search>
  <BROWSER
-    update="http://mycroft.mozdev.org/plugins/Vivisimo.src"
-    updateIcon="http://mycroft.mozdev.org/plugins/Vivisimo.png"
+    update="http://vivisimo.com/mozbar"
+    updateIcon="http://vivisimo.com/images/Vivisimo.png"
      updateCheckDays="7"
  >

Ok, so time to go to the detailed stats as usual.

==============================================
General summary
   From                  05.01.2004 00:01:37
   To                    11.01.2004 23:54:41
   Duration                        167:53:04
   Searches/Hour         153.30

> General summary
>   From                  29.12.2003 00:00:25
>   To                    04.01.2004 23:59:57
>   Duration                        167:59:32
>   Searches/Hour         126.48

Last week already, we had a huge increase in searches/hour, this time 
again, however not as strong as it was last week (+38 per hour), this 
time "only" +27)

Total Searches         25736    100%
   0 result              1521      6%
   1-15 results          5128     20%
   16-30 results         2099      8%
   31+ results           4216     16%
   QuickLink            12767     50%

> Total Searches         21247    100%
>   0 result              1023      5%
>   1-15 results          4136     19%
>   16-30 results         1525      7%
>   31+ results           3543     17%
>   QuickLink            11020     52%

Looks like things are stabilizing here. Last week we had a major change 
du to the introduction of categories.

Advanced form           8038     31%    100%
   Nothing special        163      1%      2%
   Country                 63      0%      1%
   Language               192      1%      2%
   Category              7620     30%     95%

> Advanced form           6626     31%    100%
>   Nothing special        156      1%      2%
>   Country                 44      0%      1%
>   Language               142      1%      2%
>   Category              6284     30%     95%

No change, other than the absolute numbers...

Category Top 10         7621
   1.  Computer           785            10%      .
   2.  Language dictionary735            10%     +1
   3.  File Sharing       650             9%     -1
   4.  Programming        542             7%     +2
   5.  Academic           531             7%     -1
   6.  Reference          499             7%     -1
   7.  Music              498             7%      .
   8.  General            459             6%      .
   9.  Mozilla            403             5%      .
   10. Undefined          375             5%      .

> Category Top 10         6284
>   1.  Computer           604             10%     +1
>   2.  File Sharing       597             10%     +1
>   3.  Language dictionary556             9%      +5
>   4.  Academic           437             7%      +1
>   5.  Reference          422             7%      -4
>   6.  Programming        418             7%       .
>   7.  Music              416             7%       .
>   8.  General            391             6%      -4
>   9.  Mozilla            348             6%      -1
>   10. Undefined          301             5%       N

Surprizingly (or not?), the top 10 has remained very stable, with the 
inevitable 1-2 ranks shifts inside the top 10, given the very close numbers.

Search for a name      17788     69%    100%
   10 most frequent     13414     52%     75%
   Preprocessed           473      2%      3%
   Cache                 2330      9%     13%

> Search for a name      14612     69%    100%
>   10 most frequent     11579     54%     79%
>   Preprocessed           323      2%      2%
>   Cache                 1612      8%     11%

Nothing really interesting here. The cache has slightly increased and 
the most frequent have a smaller share than last week. Due to the 
newcomer vivisimo?

Top 75% searches       13414             75%
   1.  dictionary        2913             16%
   2.  google            2597             15%
   3.  yahoo             1428              8%
   4.  imdb              1410              8%
   5.  ebay              1045              6%
   6.  astalavista       1002              6%
   7.  amazon             866              5%
   8.  alltheweb          794              4%
   9.  altavista          774              4%
   10. dogpile            585              3%

> Top 75% searches       11029             75%
>   1.  dictionary        2527             17%
>   2.  google            2330             16%
>   3.  yahoo             1217              8%
>   4.  imdb              1100              8%
>   5.  astalavista        894              6%
>   6.  ebay               802              5%
>   7.  amazon             790              5%
>   8.  altavista          687              5%
>   9.  alltheweb          682              5%

Only interesting thing here is that this time, 10 searches made it to a 
total of 75%, while last week only 9 searches could achieve that.

Top 3 bad query strings
   1.  vivismo             21   0.08%   1.38%
   2.  dhl                 11   0.04%   0.72%
   3.  phone               10   0.04%   0.66%

> Top 3 bad query strings
>   1.  shockwave           12   0.06%   1.17%
>   2.  gamefaqs             9   0.04%   0.88%
>   3.  ask.com              8   0.04%   0.78%
>
>> Top 3 bad query strings
>>   1.  metager             10   0.07%   0.87%
>>   2.  gamefaqs             8   0.05%   0.70%
>>   3.  mycroft              7   0.05%   0.61%
>>
>>> Top 3 bad query strings
>>>   1.  gamefaqs            10   0.06%   0.75%
>>>   2.  crack                9   0.05%   0.67%
>>>   3.  metager              8   0.05%   0.60%
>>>
>>>> Top 3 bad query strings
>>>>   1.  google              20   0.12%   1.40%
>>>>   2.  m-w.com             12   0.07%   0.84%
>>>>   3.  google.co.uk         9   0.05%   0.63%
>>>>
>>>>> Top 3 bad query strings
>>>>>   1.  metager             14   0.08%   0.90%
>>>>>   2.  suprnova            13   0.07%   0.83%
>>>>>   3.  google.co.jp        11   0.06%   0.71%

Really interesting. Something new this time. Vivisimo topping this list 
with the nice spelling of "vivismo". There are many other misspellings 
of this, like "vivissimo", "visisimo" and so on. dhl and phone are the 
first time in this top. Let's keep an eye on them!

Top 10 for Website
   1.  dictionary                 .
   2.  google                     .
   3.  yahoo                      .
   4.  imdb                       .
   5.  ebay                      +1
   6.  astalavista               -1
   7.  amazon                     .
   8.  alltheweb                 +1
   9.  altavista                 -1
   N.  vivisimo                   N

> Top 10 for Website
>   1.  dictionary                 .
>   2.  google                     .
>   3.  yahoo                      .
>   4.  imdb                       .
>   5.  astalavista                .
>   6.  ebay                       .
>   7.  amazon                    +1
>   8.  altavista                 -1
>   9.  alltheweb                  .
>   N.  dogpile                    N

Vivisimo is really an interesting challenger. Usually, a challenger is 
waaaaay below number 10 (one fourth of the number 10), but this time, 
it's one half of the number 10, meaning the double of a regular 
challenger. The interesting thing here is: will the quicklink attract as 
many searches as it did from our unknown source at the beginning of last 
week, in which case, vivisimo could cick out altavista from the top. 
Speaking of altavista: we can see a steady fall since we started this 
top. First on rank 5 (end of November), then floating between ranks 7 
and 8 and now dropping first time down to rank number 9... On the other 
hand, the most stable in this top 10 has been imdb, which has been on 
rank number 4 for 7 weeks now!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: vivisimo-net.png
Type: image/png
Size: 9988 bytes
Desc: not available
Url : http://mozdev.org/pipermail/mycroft/attachments/20040112/48ded547/vivisimo-net.png


More information about the Mycroft mailing list