Primordial

So the Flex team has fixed the problem with Flex Builder Pro serial numbers that I wrote about yesterday. While I still haven’t heard back about the support ticket I opened yesterday, I was able to login to my Adobe account to retrieve the serial number for my purchase of Flex Builder Pro. So let the installation continue.  Be warned its still ugly…

Now that I have a serial number, I choose to download the installer from Adobe’s site.  Given that the download size is a tiny 385 MB, Adobe provides a separate download manager program that you need to use.  While I’m not a fan of download managers (I don’t like installing one-time junk on my machine), I buy that this should improve the download success rate.

The download manager launches and defaults to my Documents folder (I hate that), so I change it to save to my Downloads folder.  I then get this security message stating that my file was instead saved to my Temporary Internet Files folder.

Weird, I never see that when I download other programs.  Adobe has a technote about this on their site where they blame in on IE, but again, I don’t see this when I download other stuff, so this seems like an artifact of something out of the ordinary Adobe is doing (like using an ActiveX control to do the download??), so I still blame this confusing experience on the Flex Builder installer.

OK download is complete, let’s roll…

I run the downloaded FB3_WWEJ.exe file.  I get the standard Windows Vista security dialog asking me if I want to run this file, so I click run again.  Now the installer is up and running.  So I accept the defaults.

The first thing I notice is that the installer defaults to AdobeFlex Builder 3.0.

I hate it when programs try to install in to the root folder, but I think this might be some kind of Eclipse thing since Aptana does the same thing.

I click Next.

Oh boy, this is going nowhere good.  Why do I foresee a lot of pain in my future?

So the installer is unable to install into the default location.  This isn’t surprising since the install would need to at least elevate to admin privileges to write files there.  I wonder why I didn’t get the elevation prompt.

So I try to change the path to c:UsersspikeAdobeFlex Builder 3.0 where the installer will definitely have permission to write.

Ugh.

OK, my next guess its that the installer needs to be installed as an Administrator, so I exit the installer and re-launch the EXE by right clicking and choosing Run as Administrator.

Well, this time the installer chose a better default location. Looking better…

Alright, now we’re getting somewhere…

Crap.  I knew it was too good to be true. I’ve got plenty of disk space, and am running as admin so I can write.  So now I spend a half hour trying various techniques to get this puppy installed without erroring out.  No luck.

Let’s search for installation issues on Adobe’s forums.  None reported. 

Dead end. Let’s try re-downloading, and reinstalling.

I encounter the same problems when running as a non-elevated User, however this time it works when I run as Administrator.

Woo-hoo!  Bummer it’s installed into my Desktop folder now.  What a silly default for an installer, but there’s no way I’m changing it right now, as I need to get some actual work done today.

What! That wasn’t the actual installation?  Now its running the “real” installation.  Why on earth did that last dialog say to “Launch Adobe Flex Builder 3.0″ as if it was done with the installation?

Well, that’s a much better installation location.

Arg, the Flex Builder 3 Beta was installed into the same location, and this installer doesn’t seem to support upgrading.  (Note to Adobe, if you’re not going to build upgrade support into the installer, please install the betas into their own folders that won’t be the same as the final version).

Given my experience with this install so far, I’m very nervous about uninstalling a version that I know works.  Luckily, my experience moving from Beta2 to Beta3 was good enough to give me some faith that I won’t get totally hosed, so I uninstall.

Hmm, not sure what that’s about. Process Explorer reports the files aren’t in use, and I’m able to manually delete them after the uninstaller exits.

Oh well, moving on.  Now I tell the Flex Builder 3 installer to try again.

Now we’re actually installing.  Approximately 8 minutes of installation later, the installer completes.

Woo Hoo! for real this time??

Ugh, that message gives me great confidence.  I quickly look for the install log.

Delete Folder:            Source:com.adobe.flexbuilder.update.site
                          Status: WARNING
                          Additional Notes: WARNING - There was a problem deleting C:Program FilesAdobeFlex Builder 3com.adobe.flexbuilder.update.site

Not sure if that’s a real problem, but I’ll just assume I’m all set.

I launch Flex Builder 3 and it works!  For some reason the new version doesn’t seem to recognize the existing projects in my workspace from the beta, but that’s something I’ll look into later.

Conclusion

After 3 solid hours of installation-related work and a total of 24 hours later, I think we can finally call my installation of Flex Builder painful, but successful.  Separate from the serial number glitch, there’s a lot of room for improvement in this installation experience:

• The download manager shouldn’t do weird things that cause files to get downloaded into unexpected locations
• The first EXE unzip experience was very painful.  It failed multiple times and finally worked after re-downloading, and running as an Administrator.
• If the installer needs to elevate to administrative privileges, it should prompt to elevate.
• The first unzip experience seems like it installed the product, but it didn’t, there was yet another “real” installation after that one completed
• The installer should deal better with installing over previous versions. Either add real upgrade support, or use an installation location that won’t conflict with previous versions you have shipped.
• This installation experience is way too long long and takes way too many clicks.  Aptana is comparable in its general architecture (an Eclipse-based standalone IDE) and that installation experience feels extremely light comparatively. (Disclosure: I used to work at Aptana)

Why did I spend the time to write this blog post?  I hate, hate, hate needlessly frustrating product experiences, especially for products that I generally think are great.  Installation experiences can be surprisingly bad because product teams often consider them a secondary priority to the core product.  This can be disastrous since frustration during a user’s first experience will at least lead to a bad first impression, and can quite often lead to aborted installations and lost customers. This goes back to a belief that I’ve always had:

Product Engineers who like to do quality installation work are well worth their weight in gold.

In software it’s always necessary to weigh the cost/benefit for all issues. It’s quite possible that the worst parts of my experience occur very infrequently in the overall number of installs. If so, then I accept that the pain I incurred may not be worth the cost of making sure this never happens again.  But, if it turns out this happens a lot, it implies a software quality issue that either wasn’t caught during testing, or one that was caught, but was mis-prioritized.

Here’s hoping the experience everyone else has installing Flex Builder 3 is much better!

Wow. Adobe Flex Builder 3 went final today.  I’ve been on the beta for many months now and have liked it.  I immediately went to upgrade my copy Flex Builder 2, and what a disaster their online purchase experience is.  This is the first time I’ve purchased anything through the online store since Macromedia became Adobe.  Let me say the store merger hasn’t been kind.  Adobe creates great products, and having worked for Macromedia in the past, I know many of the engineering teams are top-notch.  But Adobe please, the let’s optimize the  purchasing system to let great products be exchanged for $$ more efficiently.

Here’s 2 hours of my life I’ll never get back…

Follow the link to upgrade my copy of Flex Builder.  99 bucks, what a deal!

Here’s my credit card… Purchased, downloaded serial numbers sent via email. Awesome. 

…Wait, now I see there’s a pro edition that wasn’t an option during my purchase.  I got standard. Crap I need pro for the charting. 

Let’s see how to call customer service to change.  The customer service links take me to a website.  Many of the links on the site are broken.

Lot’s of searching later, I have no luck finding a phone number, but their general FAQ says I can use their online system to return the product and buy the right one.

I try the return system, but it says I don’t have any products to return.  Jeez, well, maybe the system takes a while to see my purchase. 

I’ll just buy the right version anyway and work out returning the wrong version later.

Here’s my credit card… Purchased, downloaded serial numbers sent via email. Awesome.

Wait, my email doesn’t include the serial number, it says call customer service, but of course doesn’t include an number to call.  It does give me a link to contact customer service. Let’s try that.

Awesome, thanks for that.  Now I’m hosed. I already know you can’t find a phone number to call on their website. Let’s try some of the other links from the email.  Well the FAQ isn’t broken.  Ah, sweet, there’s a link to the customer service contact area. It even has a phone number!

Calling. Waiting.  Writing this blog post to focus my griping…

Waiting, Writing. Waiting, Writing.

Got a rep.  I explain the situation.  She gets it surprising well.  Needs to transfer me to the real customer service department. Gives me a direct phone# in case we’re disconnected (thank you!)

Waiting, writing.

Re-explain to everything to customer service.

“Sir, you have to wait 24 hours to cancel an order, you’ll need to call back to cancel”. Fine.  Let’s move onto the serial number for the pro edition I purchased.

“Sir, You should have received an email with your serial number”.

I got the email, but it said to contact customer service for the phone number.

“I need to create a case and may have to transfer you to the correct department.  Please wait 2-3 mins while I check on the Pro serial number.”

Waiting, writing.

“Sir, What browser are you using?” IE. What kind of Internet connection do you have? (what does that have to do with anything?)

“Sir did you get an error when checking out” No.

“Did you get an email with the serial number?” I as explained, the purchase completed, I got the serial number email, but the email says to contact customer service to get the serial number.

“Sir I need to create a case, please hold for 2-3 mins”

Waiting, Writing.

“Sir here is your case number.  You will be contacted within 48 hours about your issue.”. What the…I just need my serial number.

“Sir we don’t issue serial numbers, we are just customer service”.

OK fine whatever. Thanks.

To be continued…

Publish blog post.

I just got an ad in my Facebook news feed about a movie my friend Mike rented at Blockbuster. 

This seemed like a typical social application notification at first, so I assumed Mike was using the Blockbuster Facebook app. After looking more closely however, I realized the ad was from Facebook, not an app.  I found that very surprising!  That got me very interested in how Facebook managed to make the connection between Mike’s Facebook ID and his online Blockbuster account…

Beacon strikes

My starting assumption was that this was Facebook’s notorious "Beacon" advertising program in action, which turned out to be correct. 

Beacon launched in December and caused a big brew-ha-ha about privacy violations because the program sends notifications to your friends about purchases you have recently made.  MoveOn.org even got involved and applied pressure to make this an opt-in only program. Facebook eventually bowed to the pressure and created an privacy setting to let users opt-out of the advertising program.

http://www.nytimes.com/2007/11/30/technology/30face.html

The public outcry reminded me of the reaction I saw to the introduction of the Facebook news feed, which users eventually came to love.

http://www.time.com/time/nation/article/0,8599,1532225,00.html

http://www.insidefacebook.com/2007/11/21/beacon-concerns-like-news-feed-concerns-of-a-year-ago-will-fade/

When this first blew up, my assumption was that things I buy through Facebook were broadcast to my friends (for example: I follow an ad link from Facebook to buy something on Amazon, and then Facebook sends an alert to my friends).  This struck as the cost of using a service that notifies your friends about most of what you do on the service.  No biggie, quit your whining.

My understanding of Beacon turned out to be completely wrong and now that I understand the way it really works, I can only say it is absolutely genius and quite possibly evil.

The investigation (IM-style)

spike: low pri question for u when you get a sec
mike: shoot
spike: Facebook is running an ad about the fact you added 24 to your blockbuster queue
spike: What activity caused that?
mike: that’s a good question
mike: on the blockbuster side, it must be linked to the blockbuster movies by mail system
mike: netflix competitor
spike: Actually, it says you added hitman to your queue
mike: but for facebook to be linked to blockbuster.com
spike: then gave me and ad for 24
mike: it’s listing what I added?
mike: personally
spike: that’s what it says
spike: "Mike Nimer added Hitman to their queue on Blockbuster"
mike: that’s really annoying, I don’t want that data shared
mike: and I’m pretty sure I haven’t added a blockbuster facebook app
mike: to my account
spike: I’ll tell u how to disable, but I’m trying to figure out how they know
spike: do you use Blockbuster?
mike: I would imagine that I have some blockbuster app
spike: that’s what I was wondering.
mike: on my account, which pings blockbuster and gets my last video
mike: although it doesn’t work very well, I added hitman weeks ago
mike: that’s the only way it could happen, right?
spike: Check out your apps list when u get a sec to see if you have Blockbuster installed
spike: yes, somehow they have bound your Blockbuster account to your FB ID
mike: ok this is wierd, I don’t have it listed
spike: think you did in the past?
mike: no
mike: I’m pretty sure I wouldn’t of bothered to add it
spike: hmm
mike: when did it show up that I added hitman, is it a new post?
spike: crazy
spike: It showed up in my news feed.
spike: FB generally inserts 1 ad per day in there
spike: I know they had an ad service recently that they got bashed for cross advertising personal info with
mike: but for blockbuster to know about me, and know I’m linked to you
spike: this is exactly what it looks like, but I was interested in how they linked it
mike: that kind of scares me, granted blockbuster itself is harmless - but in general the idea that it can work this way
spike: They also had some kind of connection with Amazon where they could tell your friends about stuff you bought
mike: doesn’t the facebook api allow you to find your firiends
spike: yes. But this ad is clearly coming from Facebook
spike: but Blockbuster gave them the info
spike: its not app spam like you would do if it was implemented directly by the app
mike: maybe it’s some deal with facebook
mike: could the social ads be it
mike: Privacy Settings for Advertisements Facebook occasionally pairs advertisements with relevant social actions from a user’s friends to create Social Ads. Social Ads make advertisements more interesting and more tailored to you and your friends. These respect all privacy rules. You may opt out of appearing in your friends’ Social Ads below.
mike: this must be it
spike: yeah. that’s how you opt out of "Beacon"
mike: I mean it let’s me set "Appear in Social Ads for"
mike: and considering that I added hitman weeks ago, which is probably when they started doing the data mining for the ad
spike: Here is the app: http://www.facebook.com/blockbuster
mike: I’ll bet this is related
spike: click on it and see if you’re a member
mike: I’m not
spike: here’s controversy article: http://www.computerworld.com/action/article.do?command=viewArticleBasic&articleId=9053002  
mike: I just changed my social ad option. let me know if you ever see this again
mike: if so I’m going to have to go nuts on facebook
spike: where did you find the opt out?
spike: I’m just facinated that they made the connection btwn the 2 sites
spike: I wonder if its based on your email
mike: under privacy
spike: if so, that’s just Fn awesomely evil
mike: keep that news post
mike: I think so
mike: actually I don’t think it’s email
mike: facebook, is linked to my work email
mike: and blockbuster is linked to my @yahoo email
spike: hmm.
spike: F’ me!
spike: "Facebook Beacon works through the use of a 1×1 GIF web bug on the third-party site and Facebook cookies."
spike: jeezuz those guys are smart
mike: wow
spike: http://en.wikipedia.org/wiki/Beacon_(Facebook)  
mike: that’s what it was
spike: Yeah I knew it was, but I was trying to figure out how they do it
spike: now I know. Wow
mike: that’s evil
mike: that’s a spammer trick
spike: One of the big customer advocate groups went apeshit about it when they launched it
spike: that’s where the opt-out option came from
mike: I remember that
mike: I guess you still default it on
spike: I thought they had changed it to default to opt-out
mike: that’s what the articles claim
mike: but I barely use facebook, I have the most basic account
spike: MoveOn.org fought it hard
mike: I’m glad to hear that. what I’m curious to hear more about is what the computerworld article talks about
mike: with the supreme court hearing and blockbuster
mike: this sure seems like a violation of that
mike: most companies can plead ignorance, but not blockbuster
spike: yeah
spike: Do you have a message like this in your feed? http://bits.blogs.nytimes.com/2007/11/29/the-evolution-of-facebooks-beacon/  
mike: I don’t see anything like this on facebook, checking blockbuster now
spike: should be on FB
mike: don’t see it, but found this on blockbuster
mike: http://www.blockbuster.com/help/faq/  
spike: Do you see anything about blockbuster on this link?
spike: http://www.facebook.com/privacy.php?view=unconfirmed_actions  
mike: yes
mike: and I’ve now turned everything off
spike: can you send me a screenshot of what you see
spike: I’m writing a blog post about all this
mike: it’s not that facebook does this, it is that it’s on by default
spike: agree
mike: in the words of google, that is evil

So, what’s it all mean?

It turns out that cookies set by Facebook are being used by other sites that you visit to link your activity on the site back to your Facebook account.  Facebook then uses that activity information to send ads to your friends.

Dave McClure gives a good rundown of the total experience (though I think the experience has changed a bit since his post).

http://500hats.typepad.com/500blogs/2007/11/facebook-beacon.html

I’ve been involved in Web application programming a long time and I’ve always mocked the folks who disable browser cookies by calling them cookie-freaks because I felt the good of stateful web applications greatly outweighed the evils. This is the first time I’ve really encountered a use of those cookies that really made me feel like I’m being watched. I now feel a bit oblivious, because it turns out this cookie-sniffing technique has probably been used by big portals like Microsoft, Google, Yahoo for a long time to help optimize ads they serve you.

I have to admit, I love a good hack, and this one is awesome.  Now I just have to let it sink in to see how I really feel about all this and if there are other good ways to use this technique.

Is it evil? Where’s Darwin when you need him?

I’m not convinced this kind of tracking and advertising is evil, but I do think this is the kind of thing people should get to opt-in to instead.

It turns out Facebook already has a great opt-in mechanism that makes sense for this, its called Facebook applications, and when you add them to your profile, its because you are interested in sharing information about the services the application provides with your friends. Facebook even provides a huge warning dialog when you add applications that lets you know that you are letting the application have access to your friends and your personal information. And *everyone* on Facebook is well aware that these apps send notifications about your activity to your friends.

I actually like the idea of implicitly sharing some of my outside Facebook activity.  Some of the sites I visit say a lot about me, and I want my Facebook profile page to be a reflection of me and my interests.  For instance, I like to have the music I’m addicted to on my profile page, so I created an app to broadcast songs I’m playing on my Zune.  And since I’m a movie buff, I’d like to have a Netflix application on my profile that shows the movies I’ve recently rated so my friends can see what I thought of them.  Hell, I’d also like to put my Amazon wishlist on my profile so folks can know about things I’m interested in buying.

However, its important to realize that just because I use a service, that doesn’t mean its a good reflection of who I am, so I don’t think Facebook should just inject that info into my profile. 

This comes back to the original problem. If Mike thought that his movie rentals were a good reflection of his personality and wanted to share info about movies he is renting, he should have added the Blockbuster application to his profile and then Blockbuster could link his account to facebook and inject ads and notifications into his news feed.  If he finds these notifications are annoying or too personal, then he can remove the application. This forces Blockbuster to make their application and the notifications they serve appealing to FB users.  If the app is flexible enough to make its content/notifications/ads a good reflection of Mike’s personality, then he’ll keep it, otherwise he’ll uninstall it.

Opt-in for good apps/ads, opt-out for bad ones - Darwin at its best. Beautiful.

What should FB do?

Installed applications already have the ability to track users and send activity notifications to friends, which can include ads. This is basically the same thing that Beacon does.  The advantage of Beacon is that advertisers can get around the problem of making their applications interesting enough to get users to install their applications by paying Facebook for the privilege of injecting application-like notifications into the user’s activity stream without consent.

This seems backwards. What Facebook needs right now is great applications that keep users engaged.  The fact is, most of these partners probably have activity streams that users would actually like to share with their friends.  Facebook should be helping these partners create compelling applications, and then they should help promote these apps so they get installed by lots of users (with consent). Facebook should then move its Beacon program to be an advertising service offered to FB applications.   This means Facebook would need to find a way to make the use of the Beacon system a value-add for application developers (versus what applications can do on their own with notifications).  I’m not exactly sure what that value-add is, but based on what I’ve seen over the last year these FB guys are super-smart, and will figure something out.

What can you do to protect your privacy?

So I’m definitely a little creeped out by the way this Beacon thing works, and don’t really like the idea of notifications about my activity on the Web being broadcast to friends without my consent.  There are a number of things you can do if you want to protect yourself from this:

You can opt out of showing up in social ads on Facebook here:
http://www.facebook.com/privacy.php?view=platform&tab=ads

You can opt out of showing your Facebook friends what you are up to on external sites here:

http://www.facebook.com/privacy.php?view=unconfirmed_actions

[Note: this doesn't necessarily keep the site from sending Facebook notifications about what you are doing, it only prevents Facebook from showing it to your friends.

Supposedly, Facebook also shows you an alert that an advertiser is sharing information about your activity to your friends.  You can use the alerts to remove the notifications from your activity feed. I’ve never seen one of these alerts, and my friend Mike couldn’t find one related to Blockbuster on his Facebook page.

http://bits.blogs.nytimes.com/2007/11/29/the-evolution-of-facebooks-beacon/

Moveon.org also has a petition you can sign to prevent companies from telling your friends about what you buy on sites or let companies use your name to endorse their products without your permission.
http://civ.moveon.org/facebookprivacy/

Consider complaining to your service providers about sending your private information to Facebook (and possibly other advertisers).  Apparently, several of the original Beacon launch partners bowed out of the program based on the negative customer feedback. Wikipedia has a list of Beacon partners that may be sharing your information.

Supposedly, if you explicitly sign-out of Facebook, this will prevent external sites from tracking you and sending notifications to Facebook.

Finally, you can disable all your browser cookies, which is drastic, so I’ll still refer to you as an cookie-freak.

The issue with Google searches for Windows Live Writer has been resolved by a fix at Google.  Apparently, the issue was related to Windows Live Spaces in general, not just the Writer Zone blog.  The conversion of urls from the spaces.msn.com to spaces.live.com domains last month caused Google’s search engine to think something fishy was going on and caused spaces.live.com URLs to carry less weight in search results.

After the issue was picked up by Channel 9, and was then Scoblized, there was a lot of speculation about possible causes of the problem.  I’m actually very happy that the theories about this being an SEO issue turned out to be bogus.  While I can understand that SEO techniques have their uses for weighing one link better against another in generalized topic searches (say something like “WYSIWYG Blog Editor“), the idea that inadequate SEO could prevent a link as strongly tied to “Windows Live Writer” as the Writer Zone’s introductory blog post from even showing up in the results would make me seriously doubt the integrity of a search engine.

Matt Cutts from Google has provided some specific details about the fix on his blog: 

By the way, it looks like the primary issue with the Windows Live Writer blog was the large-scale migration from spaces.msn.com to spaces.live.com about a month ago. We saw so many urls suddenly showing up on spaces.live.com that it triggered a flag in our system which requires more trust in individual urls in order for them to rank (this is despite the crawl guys trying to increase our hostload thresholds and taking similar measures to make the migration go smoothly for Spaces). We cleared that flag, and things look much better now.

For a search like [windows live writer], I see the Windows Live Writer blog at number one, and the Windows Live Writer Beta product download page at number 2. Going forward, I’ll keep an eye on the spaces.msn.com to spaces.live.com migration with the crawl folks to make sure that it continues to be smooth. It also looks like Mike Torres is #1 for searches like [torres talking], so overall things look pretty good now.

I’m glad Google was so supportive in investigating and resolving the issue, and that spaces on Windows Live Spaces are now getting their proper rankings in the results.  Now, what’s up with Ask’s results?

Related posts:

• Windows Live Writer blog banned from Google
• Theories on Google results for Windows Live Writer

As the “Windows Live Writer blog banned from Google?” question makes its way around, the theories on the unexpected Writer search result rankings are starting to roll in.

Here’s a quick summary of the running theories:

• The Writer Zone is Irrelevant Theory - Google’s search result is right and the Writer Zone is actually not the most relevant link about Windows Live Writer.  There have been several references to Google weighing the <title> element very heavily, so since the title of the blog isn’t “Windows Live Writer”, it doesn’t get ranked as highly as all of the other pages that do.
• Link-bomb Theory - The sudden and dramatic increase in the number of inbound links to the Writer blog when the product launched triggered some kind of link-bomb safeguard.  This safeguard caused the search engine to exclude the detected link-bomb target in search results.
• Trashy HTML Theory - The HTML generated by Windows Live Spaces is so bad that Google’s search indexer won’t include it.
• Google is Evil Theory - Google is mean and forcefully removed the Writer blog.
• Blogs are Hot Air Theory - Google’s algorithm weighs inbound links from blogs very lightly.  Since most links to Writer Zone are from blogs, this makes the Writer Zone ranking more volatile as more non-blog sources link to other websites (like the Writer download URL or http://ideas.live.com)

I don’t subscribe to the Writer Zone is Irrelevant Theory since the Writer blog was the top link for the first week after the product launched, and then suddenly disappeared completely from the results. We’ve been diligently monitoring all references to Windows Live Writer since the product launched and it seems to me that a very large percentage of them include a link back to the Writer Zone blog. A quick search on Technorati will give you some idea about the comparative relevance (at least measured in inbound links) of the Writer blog versus the current top search result:

• Links to Writer Zone
• Links to the Writer installer (the current top search result)

I don’t subscribe to the Trashy HTML Theory because the HTML generated by Spaces is not that bad. Any good HTML parser can deal with far worse.

I already weighed in on the Google is Evil Theory, and I think its bogus.  Call me naive, but it’s silly to think Google would waste its time (and more importantly money!) on adjusting its search rankings to drop the Writer Blog.  Also, given that Writer is a tool used by bloggers, its not worth the risk of the firestorm it would cause if they ever got caught.

I’m most intrigued by the Link-bomb Theory.  It makes a lot of sense that Google would have safeguards in place to catch attempts to influence page ranks.  Google has done good work to prevent comment spammers from influencing page ranks. I don’t know what kind of link-bomb detectors they have in place, but if there are any…then just like junk mail filters sometimes flag real mail as junk, its easy to see how a flood of new links to the Writer blog could be seen as a link-bomb.

I’d like to add that I have a contact at Google who is being very responsive about investigating this, and it looks like Scoble is going to ask around on his visit to Google tomorrow too.  So I think we’ll soon have some kind of answer to the mystery.  I’ll post the final answer when we know.

Whatever the answer, I look forward to finding a way for our customers to get an informative overview when searching for the product rather than the “File Download - Security Warning” dialog they are currently getting when clicking the first result.

For almost a week now, I’ve noticed that searching for “Windows Live Writer” on Google no longer brings up the Writer Zone blog on Windows Live Spaces.  I know for a fact that there are way more inbound links to the Writer Zone blog in reference to Windows Live Writer than to anything else.

Here’s a sampling of various search engine results for “Windows Live Writer”.

• Google search
• Windows Live search
• Yahoo search
• Ask.com search (Ask’s results are as strange as Google’s)

I’m having a devil of a time getting Google to return the link to the Windows Live Writer blog in almost any search.  Here’s a smattering searches that pass and fail to bring up the Writer Zone blog:

• “Windows Live Writer” blog - fail
• Introducing Windows Live Writer - fail
• windowslivewriter - success
• “Windows Live Writer” Zone - fail
• “Welcome to the Windows Live Writer team blog” - fail
• windowslivewriter.spaces.live.com - success

Now, I’m no conspiracy theorist, so I don’t really believe Google yanked our Writer blog on purpose.  I would certainly never accuse them of trying to bury Writer just because they recently re-announced Writely :-). However, it definitely seems like something happened that caused Writer’s blog to get de-emphasized in their rankings.  If anyone can figure out how that happened, you may have discovered a powerful weapon for de-listing your competitors in Google search results.

Update: Check out the latest theories on Google results for Windows Live Writer.

Update 2: Google searches for Windows Live Writer are fixed!.  I’m happy to report that all of the ”Windows Live Writer” searches linked above are now successfully returning the Writer Zone blog as the top result.

Dear “Worst episode ever”,

Posting comments on your blog is generating permission denied errors, so I guess I’ll have to comment via a trackback. I hope those are working better

I’m glad to hear you got Windows Live Writer working. I’ve encountered a few other users who are having problems getting Writer working because their WordPress xmlrpc.php interface is generating bad responses. Can you tell me what you changed just so I can pass the word onto other WordPress users?

—————-

For anyone else encountering XMLRPC errors when connecting Windows Live Writer to a custom WordPress installation, we’ve found a few instances where WordPress plugins installed on the server where causing invalid XML-RPC responses to be generated.  If you are having problems, try disabling all of your WordPress plugins, and see if Writer can connect.  If this fixes Writer, you can try re-enabling each plugin one at a time to find the one that is breaking your XMLRPC interface.  If you find a WordPress plugin that is breaking Writer’s ability to connect, please post a comment here so I can follow-up with the plugin developers to resolve the problem.

Also, any WordPress experts out there with ideas about how to solve Michael’s WordPress problem, please help.

As the dust is settling from our launch of Windows Live Writer this week, I’m starting to see a lot of these Writer temporary posts falling out of the blogosphere.

Marc Orchant recently posted a funny (but sadly true) comment about Writer being a litter bug, and Anil Dash commented that it provides a fun way to see evidence of users trying out Writer. Rick Segal is more annoyed and less delicately refers to them as “turds”.

If you don’t know what I’m talking about, take a look:

• Technorati Search: Temporary Post Used For Style Detection

So what exactly is this temporary post?

One of the really cool features about Windows Live Writer is its ability to edit your blog posts in the same styles that are used on your blog.  In general, this provides a much better WYSIWYG authoring experience since you get a more realistic presentation of what your post will actually look like when it is published.  This makes dealing with the laying out of text and images much easier.

As part of the process of configuring Writer to connect to your blog, Writer creates this temporary post so that it can detect the style-related HTML and CSS associated with your blog posts.  After the style detection is completed, Writer attempts to delete the temporary blog post.  If the blog server responds with a success message, then Writer assumes the post was deleted, and continues with setting up the blog.  If Writer receives a failure message from the blog server, then Writer assumes that the deletion operation was unsuccessful and pops up a warning dialog to let the user know that the temporary post is still on the blog and needs to be manually deleted.

So what’s going wrong?

When designing this style-detection feature, we expected there would be some cases where the deletion of the temporary post would fail and wanted to make sure users were clearly warned about the fact that the temporary post was still on their blog.  We also figured there would be some isolated cases where a blog server would respond with a message acknowledging the post had been successfully deleted, but then fail to actually perform the deletion correctly. Unfortunately, the rate of failure for this latter case is higher than we expected to encounter based on testing we did before Writer was released.

We have identified a few ideas about what is causing the problem:

The delayed file-generation effect

Any user who has used Blogger or MovableType is very familiar with the process of regenerating their site when a new post is added.  This is necessary because these blog servers actually update/generate a set of static files for each page of the blog.  Whenever a change is made to the blog (adding/modifying/deleting an entry, changing the blog’s theme, etc), the blog server needs to execute an update of the files that are affected by this change.  In working with customers on this temporary post issue, we have encountered numerous cases where the temporary post was successfully deleted from the blog server’s internal list of entries, but the cached set of static files had not been regenerated.  In those cases, the temporary post will stay on the site until another operation occurs that forces the server to regenerate the site (like publishing a new entry).

This particular manifestation of the problem has been very confusing for users because they see the temporary post on their blog’s main page, but when they go to delete the entry using their blog admin interface (or using the Writer open post dialog), they don’t see the temporary post listed.  To resolve this issue, you just need to login to your blog admin, and regenerate your site.  Since the temporary post is no longer in the list of posts known to the server, it will not be included in the regenerated site.

The ping effect

Ping servers provide a nice mechanism for telling search engines (and other interested services) that a blog has been updated.  When a blog is updated, some blog servers have support for automatically sending a ping notification to these ping servers so they can examine the updated blog content.  Search engines like Technorati will use the ping broadcast as a hint that they should update their search indexes to include the new content that has been posted to your blog.  If your blog is hosted on blog server that has any kind of delay between the temporary post being deleted from the main site (like the ”delayed file generation effect” described above), the ping effect will increase the changes that your temporary post will get indexed by a search engine. 

The FeedBurner effect

FeedBurner is a very cool service that will act as a proxy for your blog’s RSS feed so that it can gather and report statistics about the usage patterns of your feed.  I don’t have exact insights into the way FeedBurner works, but in trying to simulate some of the temporary post scenarios, it seems like FeedBurner will cache your RSS feed until it detects that a new entry has appeared in your feed.  In my ad-hoc experimentation, I found that if a request was made for my FeedBurner feed while Writer was processing my blog styles (meaning the temporary post was still present), FeedBurner actually managed to detect and cache the RSS entry about my temporary post in its RSS feed.  Unfortunately, after the temporary post was then deleted, the FeedBurner RSS feed still contained the entry for the temporary post.  The bummer here is that this creates a window of time where the temporary post can then appear in the RSS feed even though it no longer exists on the weblog.  Again, my ad-hoc experimentation seemed to indicate that this temporary post would remain in my feed until the next new post was made to my blog.  FeedBurner is just one example of a service taking advantage of a blog’s RSS feed that might cache feed data and inadvertently propagate these temporary posts after they have been deleted. This caching behavior is generally not a problem for normal blog usage, however it creates an unfortunate amplification effect for these temporary posts the Writer is using for style detection.

So what’s the plan going forward?

We have been following up with users who have expressed concerns about these temporary posts, but unfortunately, we don’t have a better approach up our sleeves for automatic style detection at this time. 

For the next beta, we plan to continue improving the robustness of our detection of orphaned temporary posts so that we can more accurately notify users if the posts are still visible on their blogs after a successful deletion. We are also planning to include a prompt in the blog setup wizard before performing the style detection to allow users to opt-out of style detection (and therefore avoid the temporary post). Finally, we will continue digging into the most common environments where this problem is occurring to see what we can do reduce the occurrence.

Given the wide range of hosted and custom-installed blogging engines in use these days, it’s unrealistic to expect that this will ever be completely resolved as long as it is necessary for Writer to make a temporary post in order to accurately detect the styles of a blog.  It would be great if, over time, we could help with new initiatives for evolving the discovery and publishing hooks exposed in blogging engines for tools like Writer and BlogJet to use to improve end-user authoring experiences and reduce these types of unfortunate annoyances.

That concludes this detailed lesson in the mysterious temporary posts. Anyone with further insights, suggestions, or questions, please feel free to let me know here or on the Writer blog.

 Other blogs commenting on Writer temporary posts

• LaraSoft: Windows Live Writer Beta Launches