Watson

re: Watson

stefan demetz — Wed, 04 Feb 2004 10:04:00 GMT

Is there a webservice for it?

re: Watson

moo — Wed, 04 Feb 2004 10:08:00 GMT

Watson is great for saving time. The problem is, how does a developer get access to those buckets? Its not exactly open to developers is it.

It would also be nice to have the RAID database update the occurance count from watson hits. This would help prioritize issues.

Watson is limited to internal and limited external, say Joe bloggs wants to write an application he cannot get access to the Watson buckets so in effect the technology is pointless outside Microsoft's use.

re: Watson

Scott Galloway — Wed, 04 Feb 2004 10:21:00 GMT

It would be fantastic to have an application block which offered this functionality - imagine that every bug in every piece of software you write is instantly known and reported back. So, how about it, a .NET application block which uses a web service to report crashes in .NET applications...

re: Watson

moo — Wed, 04 Feb 2004 10:21:00 GMT

I just tried to add to the blogg comments using Post reply in RSS Bandit , doesn't seem to work.

Anyway, I was saying that my favorite automated test is to generate the data from a template with a bit of randomness in that template (I use an XML template with placeholders for literal values, generaeted data and random data). Its very effective in some problem domains as I have demonstrated to some people. I think I successfully polverised a product and have proved that EVERY similar product is vunerable to the same technique.

re: Watson

moo — Wed, 04 Feb 2004 10:31:00 GMT

One thing I have seen (and this would be culture dependant) is that in one country I worked in people where horrified that I did something I shouldnt have on a web form (and this crashed theyre IIS server) because we are not suppost to do it that way. You can imagine my reaction to this stupidity. It was like no no you are not suppost to find bugs by doing it the wrong way, only do it the way you are suppost to use it. I think at the end of the day, they just started to believe that things are not always so rosey.

My simply reply to people who think like that is "would you rather we find the bugs inside these walls before we ship, or would you rather I find them outside these walls and publically announce them with repro scenarios on zero day - which I would do) because do not think that somebody out there will care how you should or should not use it, its the result that counts".

They changed theyre tune after, the more they start to nit pick, the more I would target a specific area. Talk about putting ones tail between theyre legs and facing reality.

Another thing a SDE/T should do is run the code through analyzers and look at possible "mr.fancy pants code" and target that (randomness values do a great job for hitting very hard to get paths).

If I see a developer act big and all mr.fancy pants ego, I go in and totally destroy theyre code. I hit hard and i bite back. I wont stop until a product is hard to break.

I dont belive in pussy footing around when it comes to breaking software, and neither will a person outside those walls with the intent of doing the same.

re: Watson

Ivan Towlson — Wed, 04 Feb 2004 11:09:00 GMT

Of course, there is a limitation in that this stuff only picks up fatal crashes, not misbehaviours. Suppose one person in a million encounters a bug with saving certain documents -- a really nasty one where it keeps losing user data in the saved documents. Most victims of such a bug would perceive the product as highly unstable, far worse than the odd crash once in a while (especially with autosave!); but because the *process* is still running, Watson never hears about it, and the developers don't get that valuable one-in-a-million repro data.

This is obviously a much harder problem: it's easy to automatically detect a process crash, but there's no obvious way to automatically realise that the program is doing the wrong thing. And I don't mean to deprecate the efforts of the Office team in improving that baseline stability. But it does illustrate that stability from the *user* point of view is much more than the low-level issue of the process staying up.

re: Watson

moo — Wed, 04 Feb 2004 11:16:00 GMT

Give all your testers at least 2 way machines for concurrency or at least give them access to such configurations.

What strikes me odd is you develope on Dell machines yet sell Compaq configurations all because of contract obligations. I guess thats management for you, cranial rectosis syndrome.

re: Watson

moo — Wed, 04 Feb 2004 11:26:00 GMT

You spout on about "Trustworthy computing" yet your employees cannot even be trusted (usually contractors).

Take for example...

http://www.internalmemos.com/memos/memodetails.php?memo_id=1664

http://www.internalmemos.com/memos/memodetails.php?memo_id=52

http://www.internalmemos.com/memos/memodetails.php?memo_id=65

http://www.internalmemos.com/memos/memodetails.php?memo_id=92

http://www.internalmemos.com/memos/index.php?search=Microsoft&x=0&y=0

Thats just a few.

Not including the easter eggs, back doors left in by developers etc. Ive seen this for myself so I don't buy your FUD.

re: Watson

Dennis — Wed, 04 Feb 2004 15:02:00 GMT

Scott, I'm doing something like that. Pretty simple, it's just a webservice that's called anytime an exception is thrown, passing up the message and call stack. Just be sure to put a try block in that exception handler...

re: Watson

Christopher Coulter — Wed, 04 Feb 2004 15:59:00 GMT

No easy answers. Somewhat the concept of software engineering itself, one cannot interpolate. But interesting logic. This “bug” might cause another bug to forthcome, therefore fixing all bugs be not the goal. What would be wrong with fixing BOTH bugs in your example? Not a zero-sum game, why an either/or? You have to bug-test your bug-fix too.

“...the first thing to realize is that you are not trying to get rid of every bug.”

Why not? Ignoring known defects? If other industries would do such, they get sued, product liability litigation, but in software par for course. Bugs are expected, glitches are the norm, not everything can be tested, too many variables, Ver 1.0, “we will patch these bugs later”. Sadly, thats the nature of Software Engineering.

However the “code, collect data, fix” approach, has its limits, namely spending most of the time fixing self-created problems. Great a tool to chart and catalog them, but that is still all after-the-fact. “Correct by Construction”, Component-Based Design and Perspective-Based (Peer Review) Reviewing/Testing can help greatly too. And, of course, litigation and legislation, sadly.

That being said, it must have worked wonders, at least for this version, as OneNote sure firmed up, the betas were literal nightmares. :)

re: Watson

moo — Wed, 04 Feb 2004 16:03:00 GMT

Thats the problem with software. Nobody will accept liability (Except IBM on AIX and the big iron but not Linux).

Everybody is so quick to point fingers and play the blame game and quick to say "we are not responsible" yet they made the product.

Accountability is something that is missing from software. This would be more so for software only products, and less of the case for "process control systems" that have software as an integral part.

Its about risk taking.

Another problem with software is the amount of immature people in design, all too quick to prove ego and funky code blocks and less on good solid coding. All to eager to show this obscufated line of code that looks cool but most likely breaks when any granny farts within a 1 mile radius - Ive seen this attitude all too often and as a result of alot of people fresh out of University with no clue. Its even worse on the opensource developement model. Show me preplanning and modelling on alot of OSS projects - its nowhere.

re: Watson

Don — Wed, 04 Feb 2004 16:13:00 GMT

You wrote "Microsoft offers this data freely for anyone writing applications on Windows. You can find out if your shareware application is crashing or its process is being killed by your users often." Could you please post something on how external dev's get access to that data? I'm not finding the right google keywords to pull it up.

thanks in advance,
-Don

re: Watson

moo — Wed, 04 Feb 2004 16:25:00 GMT

Would be nice if we can configure watson to send our application dumps to a server of OUR CHOSING. Maybe thats why there are other solutions out there that do allow this.

Found it

Don — Wed, 04 Feb 2004 17:32:00 GMT

In case anyone else is interested in signing up for Watson reporting info, go to https://winqual.microsoft.com/info/default.aspx and click on the "Windows Error Reporting" links.

-Don

re: Watson

moo — Wed, 04 Feb 2004 17:36:00 GMT

Good but why can I not configure watson to send my application to a server of my chosing?

re: Watson

Jorg — Wed, 04 Feb 2004 19:15:00 GMT

The last bug isn't fixed until the last user is dead.

re: Watson

Hi — Wed, 04 Feb 2004 23:14:00 GMT

Just a user, piping up to say that I think Watson is cool in theory, but incredibly annoying in practice: when there's a problem, its dialogue box refuses to go away no matter what I click. I have to kill all the processes to shut it down. Is Watson reporting its own problems when that happens? Who watches the watchers?

Chris Pratley has a gem

public MattBerther : ISerializable — Thu, 05 Feb 2004 02:46:00 GMT

Chris Pratley has a gem of a post today discussing how Microsoft is dealing with bugs. His post is an excellent foray into why bugs are so hard to track down and the steps they are taking to resolve them....

re: Watson

Rob Mensching — Thu, 05 Feb 2004 05:41:00 GMT

moo,

You can configure Watson to report where you want to via Corporate Error Reporting: http://www.microsoft.com/resources/satech/cer/

also, just FYI, writing easter eggs and/or backdoors are well known by all Microsoft developers to be "termination offenses". If you find an easter egg in any product released in the last two years and report it to Microsoft, I am positive the code will be removed (if at all possible) and the developer(s) that checked the code in will have a very stern talking to (which may include being shown to the door).

re: Watson

Chris_Pratley — Thu, 05 Feb 2004 08:21:00 GMT

Wow, what a lot of feedback. I wrote a whole post to respond in detail. Here are some quick comments.
1. Thanks Don, for posting that link, and you too Rob.
2. I am not sure I understand the "FUD" comment, given what that term actually means.
3. Yes, the Watson process can sometimes take a long time to build the dump - you might think it has hung. I've never seen it actually hang, but it can look like it at times if you have an app crash that has a truly enormous working set.
4. All code has bugs, no matter what methodology you use, so Watson-like techniques are going to be useful even if you have perfectly planned and architected code.
5. Watson in the future will be modified to collect information on bugs beyond crashes and hangs, and incomplete setups. Stay tuned.

re: Watson

moo — Thu, 05 Feb 2004 08:54:00 GMT

ICSA - Known back doors and easter eggs. I think found by the Telcos (even more embarassing when found by customers). Probably now removed by QFE.

re: Watson

moo — Thu, 05 Feb 2004 08:54:00 GMT

Im not denying the usefullness of Watson, I have used this alot and know the huge gains of doing so, just I was unsure if this could be used by myself for small projects free?

re: Watson

moo — Thu, 05 Feb 2004 10:37:00 GMT

The reference to "Trustworthy computing" was because its a case of "Do as we say, not as we do". Regularly we have leaks of internal memos and not just from contractors but also apparently to FTEs. Its a question of trust and yes there is a better control on the content of the products code and less backdoors and easter eggs but even recently there have been cases (ICSA being the most recent I can recall - and thats the ones we know about).

Then having seen alot of code, the quality leaves much to be desired, lack of comments, deleberate obscufaction to hinder managability and readability by some developers (Ive seen this and they admitted doing it intentionally).

Which part of this should we trust?

Actions speak louder than words.

re: Watson

moo — Thu, 05 Feb 2004 11:00:00 GMT

What also about the word thesaurs with racist word suggestions and the icons with other symbols, did the developer who added those have prior permission to do so?

Trust is something that is earned, so far its not a good track record. Its not about the bugs and exploits thats a fact of life in software, its about intent.

re: Moo

Viper — Thu, 05 Feb 2004 17:46:00 GMT

Hey Moo,

Why don't you keep the conversation to the topic at hand? Chris is writing a helpful article and you are off in the weeds criticizing something 100% unrelated. I am pretty sure Chris isn't in charge of all employees at Microsoft. He isn't in a position to code review every product. He can't control rogue employees/contractors who get some perverse sense of pride in forwarding confidential information. I realize you must like hearing the sound of your own voice, but please at least warble on topic.

Chris Pratley on bug fixing

Niall Kennedy's Weblog — Thu, 05 Feb 2004 23:55:00 GMT

Chris Pratley writes about bug fixing within Microsoft and how Watson, a software reporting tool, has changed their debugging process. He follows up with a second post about attention to detail and how hardware obscurities sometimes lead to buggy software....

re: Watson

Curt — Thu, 05 Feb 2004 23:31:00 GMT

Goodness, I wasn't at all aware of Watson/WER. This is a fine service to offer! Now, is there something similar for managed applications? I see that ReportFault() takes LPEXCEPTION_POINTERS, which isn't generally the kind of thing I have access to in managed apps....

Nice writing, I hope you keep it up.

Curt

re: Watson

moo — Fri, 06 Feb 2004 07:30:00 GMT

Actually it was on topic, he started to talk about "Trustworthy Computing" if you read. And I quote.

"This is an example of what BillG calls "Trustworthy Computing". You may think that's a bunch of hoopla, but in many ways it is quite real - things like Watson are used to directly improve the quality of our products."

Which part of this is NOT on topic?

re: Watson

Udi Dahan - The Software Simplist — Fri, 06 Feb 2004 21:46:00 GMT

In my systems, I make sure that any exception that isn't handled is logged, and/or sent to me by mail. This is the absolute minimum that should be done.

When coupled with agressive pre/post condition checking and exception management you can quickly hone in on most problems.

I suggest using the EIF on top of all this, so that for high quality, high speed logging of various events ( not errors ). Having a log of all important events in a system is imperative for finding those insidious bugs that creep up on the system over time.

re: Watson

SpiderJerusalem — Sat, 07 Feb 2004 00:50:00 GMT

Nice stuff :-)
Not really sure what random points Moo is making though. From what I understand any errors in dictionaries, easter eggs, etc. are ALL fixed once they come to light - that seems pretty responsible to me.

re: Watson

moo — Sat, 07 Feb 2004 11:18:00 GMT

The issue is "trust". What happened to the check-in process? Arn't they approved? What is the team lead doing. Is the developer not having the code read over by somebody on his team (another dev, tester or team lead) before checkin?

Its about trust and accountability thats all part and parcel of "Trustworthy computing".

It seems there is a breakdown of that process. I know I sure do a scan for these kind of things.

re: Watson

moo — Sat, 07 Feb 2004 11:22:00 GMT

Every issue in RAID has to be assigned from a tester to a team lead who then approves and assignes it to the developer.

Where in that raid work item, issue etc does it say "Include back door or easter egg".

I would certinally highlight these issues at a warteam if ever found and to make sure everybody knows who was responsible (or rather irresponsible for this basically what I would regard as sabotage code).

re: Watson

Chris_Pratley — Sat, 07 Feb 2004 19:54:00 GMT

Moo, our development teams do all the things you describe to try to reduce errors. However, absolute perfection is hard to achieve. With thesauri and spellers, those tools usually come from other companies who often produce a print version of a thesaurus or dictionary, and have decided to make an electronic version and license it to us. They are contractually obligated to verify that the tools do not have any issues. Of course Microsoft employees who are lexicographers also check the content, but there will be an occasionaly error. As SpiderJerusalem noted, we fix these aggressively when the errors are noticed. BTW, many of these "errors" exist in the print versions of the same thesauri and distionaries but for some reason they are only noticed when it becomes part of a Microsoft tool. This is probably because they get used more, Microsoft tools are considered by many to be the "standard", and it is entertaining to hassle us - you would agree with that part, right? :-)

re: Watson

moo — Sun, 08 Feb 2004 11:29:00 GMT

Here is an anoying bug, on EVERY IE control that ANY application is using, if I hold down CTRL key and click the left mouse button, why does IE launch?

I just had 28758475978 IE windows open up non stop with about:blank#66678 in it and it just got stuck in an infinite IE window spam mode.

There is no need for crap like this on a control.

Its probably one of the MOST annoying windows control bugs.

re: Watson

moo — Sun, 08 Feb 2004 11:38:00 GMT

Even more bugs, if i DOUBLE RIGHT CLICK on an IE windows control on ANY application it again opens IE in my face.

What genius dreamt up this?

re: Watson

moo — Sun, 08 Feb 2004 11:43:00 GMT

To reproduce this annoyance.

Open up the web page (this works for any application that uses windows controls like this text box).

In the blog Comments multiline textbox, hold down CTRL or ALT GR and click the left mouse button OR double click the RIGHT mouse button with NO keys pressed.

WTF kind of behaviour is this for a text box? Spamming IE in my face for a text box click????

re: Watson

moo — Sun, 08 Feb 2004 12:03:00 GMT

User feedback from watson (non UE dumps).

Is it possible for a user to just launch Watson and just click "submit" information and it will dump the application and they can add a comment to the effect of an annoyance or bug or maybe a suggestion? I can do this with the Mozilla Feedback bug trapper thats similar to watson.

re: Watson

Chris_Pratley — Sun, 08 Feb 2004 19:37:00 GMT

Today there is no public way to run Watson and add a comment, although you can run the Application Recovery Tool which ships with Office which lets you report a hang to us for any app. This can give us a little more info than the generic Watson you get when you kill a process with Task Manager.
Internally we are experimenting with such a tool that let's the user provide comments. One problem we have is the vast numbers of comments we would receive over time make it hard to categorize them, and most people are not trained to provide comments that help us reproduce the problem. Collecting data is not really the problem - it is analyzing and being able to use the data.
If you want to provide comments on the product, suggestions, or tell us a help article was not helpful, you can do that through other channels (such as within help itself - we actively make help better every week using customer suggestions). http://office.microsoft.com has a lot of good info. You can also use the Wish site: http://register.microsoft.com/mswish/suggestion.asp

re: Watson

moo — Sun, 08 Feb 2004 20:05:00 GMT

Well, usually the people that can find the option for such features are capabable of understanding the area, thats the way it is for FullCircle's Talkback reporter that ships with Mozilla browser.

Its not very visible on the browser, actually its not, its a seperate application "Talkback.exe" so first of all to find it requires a knowledge that Joe Blogs wont have.

re: Watson

Chris_Pratley — Mon, 09 Feb 2004 06:51:00 GMT

If it is very hard to find, then you are only getting feedback from expert or technical users - not from the people who really need help with the software. Managing the bias introduced by the way you collect the data is another aspect of the problem.

re: Watson

moo — Mon, 09 Feb 2004 08:30:00 GMT

Thats why we have "Remote Assistance" on XP and longhorn, right? They can go via theyre IT department. As far as the non computer junky feedback, thats why you have usability labs, right? But seeing stuff in XP, I seriously doubt theyre usefullness of late.

Behind the times again

Matthew Schwartz — Mon, 09 Feb 2004 15:37:00 GMT

Crash reporting is an old idea and something akin to watson has existed for years on other platforms. The Gnome desktop for Linux and Unix has had this tool for at least the last five years. Every application crash can report back related details plus user comments.

You're almost claiming in your post that Microsoft invented the idea. Hardly. I, and everyone on my development team at a major financial firm, have been putting bug tracking features directly into our applications for at least 6 years. The simplest example is every time an error is raised the user can press one button to automatically e-mail the team all related info.

Just because Microsoft is finally starting to do something sensible to raise quality doesn't mean it should be hailed as a brilliant step forward. You're just barely starting to catch up with the rest of the world. If you ever finally surpass everyone in quality assurance methods I'll be the first to congratulate you.

re: Watson

Chris_Pratley — Mon, 09 Feb 2004 22:33:00 GMT

Matthew, thank you for putting the "almost" in your second paragraph. I didn't mean to imply that no one has had similar ideas. Nothing is ever truly orginal anyway if you look hard enough - any idea came from another idea, or was "independently invented" in an environment that was conducive to it being conceived. I simply wanted to write about Watson and how it improves our software - and that we provide it as a service for others too. Depending on how widely your app is used, managing the reports that come back requires some significant backend system after all. One thing that makes Watson a little different from other systems I have heard about (and I have not especially researched this area), is that it collects all feedback as long as a user lets us. Other systems tend to require that you proactively report a problem, so that you can miss segments of users who are not aware how or are comfortable with taking that step.
BTW, have you done a rigorous comparison of QA methodology in Microsoft and the "rest of the world" that backs up your assertion that Microsoft is behind everyone else in quality? We have volumes of data that indicate that our software preforms very well compared to the majority of sotware out there (from looking at the Watson data in aggregate), as well as customer satisfaction surveys etc. What similar statisticially valid data are you using? I am not being flip about this - I care deeply about software quality, but it is quite easy to toss off remarks not backed by data, so I would love to hear any data if you have it.
Most of the time comments about poor quality in Microsoft products measure the subjective frequency of comments from others around them, and do not take into account the rate of usage of our products. If 100 people use a Microsoft product and 10 are unhappy, that sounds worse than 1 person complaining about a product used by 5 people (10:1 ratio of complaint frequency), yet the complaint ratio for that product (1 in 5 or 20%) is twice the Microsoft ratio (10 in 100 or 10%).

re: Watson

Philip Rieck — Tue, 10 Feb 2004 02:50:00 GMT

Chris,
I'd love to sign up for WER, but the $400 price tag (for a Verisign ID) is a bit steep for tools that sell under $20 (or free!). While I understand why you would want to restrict this information to the author(s) of the application, I'm not sure why a Verisign ID is required -- you're trying to establish that the publisher of an app holds a secret that the requestor of WER info holds... not who in particular it is.

OneNote is impressively stable. I'd love to make all software (even and especially the $10 variety) this solid.

re: Watson

Chris_Pratley — Tue, 10 Feb 2004 05:02:00 GMT

Philip,
Naturally, *I* am not asking you for $400. Sorry, I have no idea why the WER team requires this. (I am not Microsoft, as you know :-))

re: Watson

moo — Tue, 10 Feb 2004 11:47:00 GMT

So we have to pay for this, how about making it CONFIGURABLE so when I install my application I can have it setup Watson to say, dump my bugs to a server of MY choosing. Nah, too obvious for the little guy, we dont need them, they dont add value to our platform. Bugger aff we say.

This is one developer story that sucks.

I think I have to check sourceforge for a bug trapping system I can CONFIGURE to my liking.

re: Watson

moo — Tue, 10 Feb 2004 12:29:00 GMT

The $400 goes towards the lubricant they use during theyre bitch slapping sessions in the mens loos.

re: Watson

Nathan Anderson — Thu, 12 Feb 2004 08:11:00 GMT

moo, I don't understand you. A Verisign ID costs MONEY, you are paying for that ID. Verisign seems to be kind of an internet standard for validity of a site or service or even a Wireless LAN, so now you are bashing Microsoft for USING a standard to properly verify things? I know people like you, and you will find fault with anything Microsoft does, regardless of whether what they are doing is a good thing or not.

re: Watson

Brian — Thu, 12 Feb 2004 19:40:00 GMT

I'm pretty sure that legally, Microsoft will only be able to show you Watson data that relates to your companies apps. Versign provides a univerally acceptable and managable way of validating your credentials and making sure that you only see the data you are entitled to.

re: Watson

.. — Wed, 18 Feb 2004 14:00:00 GMT

A simple login ID and password would surfice.

re: Watson

AxillIum — Tue, 24 Feb 2004 13:16:00 GMT

Not really, a login ID and password are not nearly enough.
I develop secure (online) systems for (payment) transactions and I find myself endlessly perfecting my user-verification method, since I always find a way to get around it.

Ofcourse, this also means I get better at hacking, but that's the whole idea of me working with security, finding holes and patch them.

VeriSign seems a good system, although for my applications it's not the right one.
Too bad, because for once I would like to put more effort in the rest of the application instead of the security issue (not that the rest suffers from my focus on security though).

In short: simple login ID and password wouldn't suffice.

re: Watson

please, no "moo"re — Thu, 26 Feb 2004 18:11:00 GMT

Let me see a show of hands. Who here would love to work with moo? You wont see me raising my hand anytime soon. He talks about the arrogant coder, but should probably be looking in the mirror. I bet everyone has to rub Preparation H on their palm just to shake his hand.

Trust Microsoft

Dont Think So. — Thu, 26 Feb 2004 18:13:00 GMT

Trust Microsoft with your core dumps? - I dont think so.

All your data belongs to Microsoft:
http://www.theregister.co.uk/content/6/18002.html

If they dont steal it they might just give you credit-Microsoft's credit has brought me a lot of grief:
http://www.kuro5hin.org/comments/2004/2/15/71552/7795?pid=98#162

re: Watson

JBC — Fri, 27 Feb 2004 16:28:00 GMT

While this is a noble idea, there are some very serious privacy issues with it.

I've seen the dialog, I've looked at the data and there IS personal information in there.

You explicitly state that watson information is freely available, is this just aggregate data or is access to dumps provided? If these dumps are persisted and available, can you explain to me why I would ever do anything except click "Don't Send"???

As a developer I'm accutely aware of how available information is and I work very hard to protect mine. It's a constant battle and unfortunately one we are all destined to loose, I fear I'm putting off the inevitable...

-JBC

re: Watson

Chris_Pratley — Sat, 28 Feb 2004 09:04:00 GMT

JBC, please take a look at the privacy statement included via link in the Watson dialog. We are very serious about being explicit with what exactly we are collecting and what we intend to do with it. This is an opt-in feature where we are simply asking you to help improve the product, which of course you do not have to participate in.

The crash dumps are made available only to the registered and verified creators of the application you were using, so you only need to trust them, not the whole world.

re: Watson

innocent bystander — Tue, 27 Apr 2004 23:28:00 GMT

To moo: It's not like Watson is the only choice for crash data collection on the Windows platform. You can trap all exceptions with SetUnhandledExceptionFilter and record whatever data you want & send it to a server of your choosing (using tcp/ip yourself). Of course, you'll have to do a little more yourself this way than when using Watson, but I suppose the ease of use is what you're paying for.

re: Watson

osru — Sat, 13 May 2006 12:33:04 GMT

Hi everyone! I think your site is very interesting and useful. I always bookmarked it.

re: Watson

osru — Sat, 13 May 2006 12:34:13 GMT

Hi everyone! I think your site is very interesting and useful. I always bookmarked it.

Submitting Watson Crash inforamtion

OneNote Extensibility & More.. — Thu, 25 May 2006 06:31:03 GMT

Have you ever seen a crash dialog while using a beta product (and maybe even some released products)?&nbsp;...

re: Watson

ringtones free — Sat, 10 Jun 2006 20:23:40 GMT

Thanks!!! http://www.ringtones-dir.com/get/">http://www.ringtones-dir.com/get/ ringtones site. [URL=http://www.ringtones-dir.com]ringtones download[/URL]: ringtones site, Free nokia ringtones here, Download ringtones FREE. Also [url=http://www.ringtones-dir.com]samsung ringtones[/url] From website .

re: Watson

ringtones free — Sat, 10 Jun 2006 20:23:48 GMT

Hi! http://www.ringtones-dir.com/get/ ringtones site. ringtones site, Free nokia ringtones here, Download ringtones FREE. From website .

jouer � la roulette online

bank card credit household services — Wed, 30 Jan 2008 12:30:35 GMT

In klingelt�ne kostenlos card credit download free generator

gratis polyphone klingelt�ne

prepaid visa credit card — Wed, 30 Jan 2008 13:39:15 GMT

Nur apply for student credit card casino sans t�l�chargement

payday loan fast no fax

first american cash advance — Sat, 02 Feb 2008 10:07:21 GMT

Before advance fax loan no payday advance cash loan overnight

advance cash overnight

cash fast loan online payday — Sat, 02 Feb 2008 11:10:03 GMT

Her cash loan payday till advance cash loan? payday ?

advance cash on line

advance cash settlement pre settlement cash advance — Sun, 03 Feb 2008 01:46:33 GMT

Con probabilit� roulette cash advance loan oregon