| 5:42 pm on Oct 14, 2004 (gmt 0)|
|So far, I have a luke warm reception to it as it is IE integrated only (which is evil). It also doesn't allow you to specify which directories or actual file types to index. Just not enough info coming out on this thing yet. |
I wonder - will certain firewalls block it? I know some serurity software is going to block its operation for sure.
All-n-all, it's a cool idea. I've been using Lookout for a couple months now, and I think it is just awesome.
LookOut is OE integrated only (which is evil). Sure, it searches other things as well, but clearly Microsoft screwed up when you have to use a program within a program to finally get a decent search in Windows. LookOut + Outlook feels pretty bloated too.
In the end, I have a feeling that this is another 20% of some Googler's time. This is a geek toy with minimal penetration meant to keep the Google Groupies happy. It might become big or it might become stale like the Google Deskbar.
Lastly, I have to add that I am sort of pissed off that I have the Gmail Notifier, Google Deskbar, Google Desktop, and Google Toolbar (i don't use IE but it's there), Picasa, and Hello all installed and collecting various informations about me (which I don't mind, they obviously put it to good use). The problem is that, although they all have a central contact point, Google Headquarters, they don't talk to eachother. As a matter of fact, why aren't they optionally combined in one install application?
| 5:53 pm on Oct 14, 2004 (gmt 0)|
integrating these local systray applications should be the next step I assume. Then launching a GMessenger wouldn't require a separate application manager.
I guess before that all these tools would need a clear "big picture" look, to clean up the trash from valuable
| 6:02 pm on Oct 14, 2004 (gmt 0)|
"big picture" look = Google Browser?
| 6:06 pm on Oct 14, 2004 (gmt 0)|
|integrating these local systray applications should be the next step I assume. Then launching a GMessenger wouldn't require a separate application manager.I guess before that all these tools would need a clear "big picture" look, to clean up the trash from valuable |
I guess you missed my thread, here [webmasterworld.com]? Google already has a messenger :)
| 6:11 pm on Oct 14, 2004 (gmt 0)|
I wonder if you could get this thing to run on a webserver and get a full index of your site, but limit the folders it indexes.
| 6:16 pm on Oct 14, 2004 (gmt 0)|
dvduval: if you have html files and they are all local, why not?
I wonder how much is this "little google" alike the "big google". I guess the advanced win developers will find this out quite soon.
I guess we'd have to wait for the hype to go away (at least a few days) to draw conclusions
| 6:27 pm on Oct 14, 2004 (gmt 0)|
> how are they going to make money by
> giving this thing away for free?
This aint about money - it's about survival - they must compete. Yahoo and MS are gonna give theirs away for free.
You know how good Lookout is right? Now imagine when that is tagged as a little form window right next to your start button and chained to MSN search as the default off desktop engine.
The results in Lookout are better orgainized to me, but Googles is somewhat promising. Seeing the results from my machine in what looks like a Google window is just plain creepy. I don't like that part at all.
Where is the html stored for the search results? Can we change that?
| 6:27 pm on Oct 14, 2004 (gmt 0)|
From Zdnet [news.zdnet.com...]
"The application will give Google new opportunities for advertising in the future, however. Google's Mayer said that for now, the company will not tie text ads to personal information viewed on the desktop results pages, but it may explore such avenues eventually. Google will display texts ads adjacent to search results like it typically does. A residual effect of the service, she said, could be that people use Google search more often and therefore that will boost its ad pages."
| 6:33 pm on Oct 14, 2004 (gmt 0)|
Isn't it interesting how a bunch of Linux junkies design Windows apps? :) As an aside, I was just thinking that it'd be good for their company if Sergey and Larry took Steve Job's lead and worked for a $1.00 salary, now that they're rich already. Hell, they could take that ~6 billion and stick it back in their company.
| 6:52 pm on Oct 14, 2004 (gmt 0)|
Apologies if this is obvious, but I just noticed the Desktop search keeps *multiple* cached snapshots of some docs. Sweet. It'd be nice if Google.com turned this feature on.
| 7:07 pm on Oct 14, 2004 (gmt 0)|
> Seeing the results from my machine in what looks
> like a Google window is just plain creepy. I don't
> like that part at all.
It creeps me out too. Anyone notice your desktop results appear in a regular Google search by default? If people become accustomed to using the Toolbar for desktop searches, you have a new way for additional AdWords impressions.
| 7:58 pm on Oct 14, 2004 (gmt 0)|
The only Windows machine I have access to at the moment is running NT4, so I can't look at this thing. I've got a couple of questions for you testers out there:
1. It supposedly searches the temporary internet files. Does it search Mozilla/Firefox or Opera cache files, or only those accessed with IE?
2. What does the data store look like? Is it one big file or a bunch of little ones? Text or binary? Does it look in any way potentially accessible or readable by any other application (eg. an XML-like data repository which could be parsed)?
3. Any rough idea of the correlation between the size of the data store and the size of your "My Documents" folder (where, I assume, most of your documents are)? I see they suggest 500Mb of free space in the docs...
4. Any hackers out there captured any of the data transmitted to the mother ship? Is it readable or encoded in some way?
| 8:03 pm on Oct 14, 2004 (gmt 0)|
|It also doesn't allow you to specify which directories or actual file types to index. Just not enough info coming out on this thing yet. |
From a news letter we received from Search Day (Search Engine Watch)we received this explanation for file types:
You can use some of Google's search syntax to get around this. To narrow to file type, use the filetype command. For example, this:
would bring back only email matches with the word cars mentioned. A list of known file types we've tested to date and found work are:
Word: filetype:word or filetype:doc
Excel: filetype:excel or filetype:xls
PowerPoint: filetype:powerpoint or filetype:ppt
Text: filetype:text or filetype:txt
Web History/HTML Files: filetype:web or filetype:html
Images: filetype:jpg or filetype:gif
Windows Media: filetype:wma or filetype:wmv
For email, chat and web history, you can also narrow by clicking on the count numbers, as described above. As for images, Acrobat and other files, keep in mind that only text in the file names will be matched, not any meta data or actual text contained within the files.
| 8:13 pm on Oct 14, 2004 (gmt 0)|
I agree with encyclo it's privacy invasion by default but google do come up with some good stuff
| 8:21 pm on Oct 14, 2004 (gmt 0)|
Anyone imported like the top 10 SERPs for a keyword to see if it comes back with the same results locally?
I am guessing it won't since G relies too heavily on links. But it may tell you if your on page factors are better. Assuming everything else equal (links) this may tell you who would appear first. So in sense you can maximize your on page SEO.
Let me know if anyone does this.
| 8:30 pm on Oct 14, 2004 (gmt 0)|
|Anyone imported like the top 10 SERPs for a keyword to see if it comes back with the same results locally? |
There's no way it's going to use the same algo as on the public search - for the simple reason that the data set is so different. You aren't going to spam your own desktop search, so the algo is going to be much less complex and more trusting of the document and metadata information.
Having said that, how long before someone works out how to spam the data store via their visited pages in the temporary internet store. Search for your banking and budget documents, and get back results from your IE cache for mortgage offers...
| 8:50 pm on Oct 14, 2004 (gmt 0)|
> 2. What does the data store look like?
In the Google Desktop Search folder, the following files are present:
I was looking at the .exe files for the Desktop Search and noticed a c:\src\1.0_branch\googleclient\total_recall\build\release\obj\trs\GoogleDesktopIndex.pdb - this is a Windows programs so the datastore may be a proprietary "Program DataBase" build by MS Visual Studio like VC++.
| 8:57 pm on Oct 14, 2004 (gmt 0)|
Anyone notice how Googles new desktop icon looks like the Windows Flag in XP spun into a swirl?
<added>search function is much faster than the standard searches on XP</added>
| 9:03 pm on Oct 14, 2004 (gmt 0)|
>>6hrs of indexing
Brett, it only indexes when your puter is idle for atleast 5 seconds.
| 9:06 pm on Oct 14, 2004 (gmt 0)|
Actually 7 hrs now chundru. whew.
Why would they not have a bulk index option to get it done in one shot? Mine will *never* get done. 60k files out of 2m files dones. At this rate, we are talking about 4 days to index. To just test a program I will probably use 10 times!? yeow.......
| 9:16 pm on Oct 14, 2004 (gmt 0)|
|we are talking about 4 days to index. |
Yeah, but what a job they are doing! After that you will only have your changed files updated. Stop posting and go take a nap and give them a chance to do their job :-).
| 9:20 pm on Oct 14, 2004 (gmt 0)|
|this is a Windows programs so the datastore may be a proprietary "Program DataBase" build by MS Visual Studio like VC++ |
OMG it's another .NET creation! Let's hope that it's better than Orkut! ;)
| 9:38 pm on Oct 14, 2004 (gmt 0)|
Hey Hanu (message 23), your effort wasn't entirely wasted. I thought it was funny.
| 11:09 pm on Oct 14, 2004 (gmt 0)|
As for using the desktop to game the algo. I doubt they will be the same algo since there is little link popularity through your desktop and Google knows our intentions are different when we want our PC indexed. I am not deliberatley going to try and spam myself so they can probably use simpler algorithms with fewer filters. I bet that what works for getting a high listing on your desktop search will not work for getting a high listing in Google proper.
| 11:13 pm on Oct 14, 2004 (gmt 0)|
We are now 9hrs into and only 5% done with indexing. Lookout did in under an hour. This right here is pretty much a washout for me.
| 11:25 pm on Oct 14, 2004 (gmt 0)|
I'm going to create a Mycroft search plugin for FireFox based on GDS. First, I need to know whether or not it consistently uses the same port. On my home machine, it uses 4664. Does anyone else have a different port being used?
| 11:31 pm on Oct 14, 2004 (gmt 0)|
Brett...you've been sitting at your computer for 95% of the last 9hrs. Isn't it time to take a break anyway? :)
|We are now 9hrs into and only 5% done with indexing. Lookout did in under an hour. This right here is pretty much a washout for me. |
| 12:14 am on Oct 15, 2004 (gmt 0)|
> Lookout did in under an hour. This right
> here is pretty much a washout for me.
That is hardly a fair comparison when LookOut is doing a full index hogging all processor time, while Google is gently chugging along.
The moral is for Google Desktop Search to provide a "rebuilt database in real-time" option.
You have to admit that LookOut requires .NET installation and a lot more time for the actual installation than Desktop Search.
| 12:56 am on Oct 15, 2004 (gmt 0)|
This help file [desktop.google.com] answers my first question - it doesn't add web pages viewed with Firefox to it's search. They don't even mention other browsers such as Opera, so I wouldn't hold out much hope there either.
So, to profit fully from Desktop Search, you have to be running IE as your primary browser (and use Microsoft Office for your documents and Outlook or Outlook Express for email).
sun818, what do you see if you try to open up one of those files in the data store with a text editor? Is it a text format or binary?
Finally, Google recommends allowing four applications to access the internet [desktop.google.com] without restrictions - although they don't explain why the desktop crawl program or the others should ever need to do so.
| 3:25 am on Oct 15, 2004 (gmt 0)|
> sun818, what do you see if you try to open up one
> of those files in the data store with a text editor?
> Is it a text format or binary?
It is binary. Some the files have a repeating pattern suggesting it may be fields in fixed length database.
| 6:59 am on Oct 15, 2004 (gmt 0)|
I hope that folks enjoy trying it out. In my experience, you start to notice the benefits more and more after a week or so. Things like indexing different snapshots of a document are pretty handy. encyclo, it is a different algorithm than the main web search.
If people have suggestions for improvements, the best place to give us feedback is here:
| This 95 message thread spans 4 pages: < < 95 ( 1  3 4 ) > > |