Tim Anderson’s ITWriting

I am beginning to regret my pretty permalinks. Apparently it broke the RSS feed for some subscribers. The problem was caused by a combination of factors. When I started this blog, I used a different blog engine; and when I migrated to WordPress I fixed it so that the old feed URL still worked. Moving to pretty permalinks broke that feed, although the normal WordPress feeds still worked. There was no error; the feed was just empty. The reason it broke is … something to do with redirects. Fixed now.

Apologies if this affected you – I have not been as quiet as you may have thought.

Last month I reconfigured this blog to use WordPress pretty permalinks. I didn’t rush to do this because I don’t mind the default, where the post is identified by a numeric url argument. The new permalinks do look nicer though, and some claim they help search engines, though I haven’t noticed any impact on traffic. The old links still work as well, so nothing got broken.

One advantage of the new arrangement is that awstats does a better job of reporting page views. I can now easily see which posts are the most read. I thought that might be the newer posts, but it is not. The two posts most read last month, by some margin, are both on why Outlook 2007 is slow. The next is one on trouble installing Adobe CS3. Clearly, these are very common frustrations; users hit Google and these posts have a high ranking for these particular subjects.

I value these stats. They tell me (as if I did not already know) that Microsoft has a significant problem with Outlook 2007 performance; and that Adobe needs to work on its installer for Creative Suite. Web 2.0 is great for product feedback.

Advertising is “the economic engine that powers the Web”, according to Microsoft’s Ray Ozzie. Google’s rapid ascendancy, enabled by advertising revenue, is the primary evidence for this. That said, Rick Strahl’s post on Google advertising highlights several problems with Google’s approach. It is about Adsense, the mechanism by which third-party sites (like this one) host Google advertising. When someone clicks an ad, Google gets paid, and an undisclosed percentage of the fee goes to the site owner.

Strahl runs a small business and uses Google both as advertiser and web site owner. He’s puzzled by the stats he’s gathered at both ends. As advertiser, he says that 30-40% of his hits come from link parking sites, plus another 10% which have no referrer, and reckons that these hits are worthless and in many cases possibly fraudulent. That’s up to 50% of wasted ad spend. Google tells him there is no way to opt out of link parking sites, other than by excluding specific sites; but since there are thousands of such sites and they change constantly, that is impractical.

At the other end, Strahl sees frequent deductions from the clicks on his own site, presumably on the basis that they are fraudulent or accidental (such as robot clicks). In fact, deductions from his site, which he controls and which has good, genuine content, appear to be far higher than those from the link parking sites which have no real content at all. In other words, Google seems happier to make deductions from what it pays to him, than from what he pays to Google.

He’s also curious about the ad bidding process, which always seems to end up charging him the maximum possible.

It’s possible that he has some of this wrong; but there is no way to audit Google’s figures:

In the end it feels like black magic. Google (and other advertisers as well to be fair) control the process so completely that if there’s any foul play either on Google’s part or for cheating publishers that contest clicks on the other end there’s almost no real way to tell that it’s happening and unless you have the time to keep very close tabs on it there’s no way to follow the money all the way through - on both ends. And who has that kind of time?

I find this unsurprising. The pay-per-click model has always seemed to me far too vulnerable to abuse, especially bearing in mind all those botnets. Who pays for any fraud? Not Google, but Google’s customers, the advertisers.

Some level of click fraud is inevitable, but Google’s willingness to let any old worthless bot-driven link parking site run Adsense ads is a disgrace. This stuff poisons the web, because it provides a financial incentive to post junk.

Advertisers can opt-out of Adsense, by disabling the “content network” for the ads they place. If enough advertisers do this, Google will take note.

Disclosure and to add a personal note: I am an Adsense publisher, though not an advertiser. I also use Blogads, which to my mind has a better business model for advertisers, since they specify exactly which sites they wish to use. In addition, I get to approve each ad, whereas with Adsense I have to take whatever comes. The snag is, Blogads is tiny in comparison to Google, which can seemingly always supply ads for my site.

Sometimes the Internet reminds me of Tony Hancock’s Blood Donor. You post advice when you have it, and take it back when you need it.

It was like that last night. I am following my own advice and weeding out any instances where username/password combinations are transmitted in plain text. Occasionally I send mail via Exchange as an SMTP server, so I’ve now configured this to use TLS (Transport Layer Security).

All went well until a fatal reboot produced event 32777: The LSA was unable to register its RPC interface over the TCP/IP interface. This is nasty, and causes a host of further errors which pretty much kill networking on the box. I have no idea what provoked it.

Fortunately I’ve had this before - two and a half years ago. Last time I used the blunt instrument of a repair install, but by going back to my earlier post and reading the comments I was able to apply fix this quickly:

• Change the logon of the RPC service to Local System, as a temporary fix to networking
• Make changes to local security policy (domain controller policy in this case): Add Adminstrators and Service to the Create Global Objects and Impersonate client after authentication in User Rights Assignment
• Change the logon of the RPC service back to NT AUTHORITY\Network Service

All very obscure and the kind of thing you have little chance of working out for yourself. It is all to do with changes made by Server 2003 SP1 which appear to break important stuff in some circumstances.

Why not Server 2008? All in good time.

Here are the posts that received the most comments on ITWriting.com this year:

Vista display driver takes a break (220 comments)

Outlook 2007 is slow, RSS broken (173 comments)

Annoying Word 2007 problem- can’t select text (101 comments)

Why Outlook 2007 is slow- Microsoft’s official answer (95 comments)

Adobe CS3 won’t install (35 comments)

Delphi for PHP first impressions (33 comments)

Irony: Outlook Web Access more usable than Outlook (29 comments)

Audio in Vista- more hell than heaven (25 comments)

How to speed up Vista- disable the slow slow search (24 comments)

Adobe AIR- 10 reasons to love it, 10 reasons to hate it (24 comments)

Ubuntu Desktop not used in business (21 comments)

Miguel de Icaza on ODF vs OOXML (19 comments)

Visual Studio 6 on Vista (16 comments)

Microsoft Silverlight vs Adobe Flex (16 comments)

Vista vs XP performance- some informal tests (14 comments)

Slow Outlook 2007- the comments keep coming (14 comments)

This is mostly down to Google, everyone’s favourite source of tech support. The most commented posts are about problems with Windows and Office, and reflect the number of people searching for a solution who land up on this blog. Only a tiny proportion of readers actually post a comment, so the top few posts above are evidence of a large amount of frustration.

I highly value the comments, especially when they form a reply or clarification from the organization which is the subject of the post - like this one from Zoho.

A few more stats

FireFox usage has increased from 14% in 2006 to 20% in 2007.

The biggest source of incoming links is programming.reddit.com.

The five top search keywords are: 2007, Outlook, Vista, Slow and .NET.

A bit of introspection

I enjoy doing this blog and web site, though there are a couple of frustrations. One is that I have more material than I get time to write up. Another is that while the ads on the site pay for the hosting, they don’t do much more than that, and I would like to find a way to make web self-publishing viable.

I also muse over whether the range of subjects here is too broad. I post in three broad categories:

• Software development
• Problem solving
• Anything that interests me in the tech world

Most of the subscribers to the blog probably want what is in the first category, especially as it is in this area that I can supply the most original content, sourced from interviews or conferences. The problem solving posts find a different readership via Google. My good intentions to narrow the focus more towards programming fall away when I have some other topic I want to write about, though I do keep it strictly to tech-related topics.

Update: fixed the list (missed a few)

Microsoft’s Jason Zander comments on my piece on the early history of ASP.NET:

• The CLR was actually built out of the COM+ team as an incubation starting in late 1996.  At first we called it the “Component Object Runtime” or COR.  That’s why several of the unmanaged DLL methods and environment variables in the CLR start with the Cor prefix.
• There were several language projects underway at the start.  The C++ and languages teams both had ideas (Cool was one of them), and in the CLR we wrote Simple Managed C or SMC (pronounced ’smack’).  We actually wrote the original BCL in SMC.

He says these are corrections though they seem more like supplementary information to me. I don’t have any inside knowledge of this history other than what people who should know say to me (though I do also have my own recollections of what was said publicly). He may be reacting to the idea that the CLR came out of the VB team, which Mark Anders kind-of implied.

One of the reasons I love blogging is that multiple authors can have a crack at getting the facts right. A great personal example is when I asked the question Who invented the wizard; and a good candidate came forward over a year later. If you see something inaccurate or misleadingly incomplete on this site, please do comment or let me know by email.

Kim Cameron has an amusing post on the aftermath of his blog being hacked and defaced over the weekend.

The reason for the hack: a security bug in WordPress. More proof of the problem posed by millions of apps out there on the internet with no update mechanism in place. Security fixes are made available, but not applied. WordPress has improved this somewhat by introducing an alert when you log-in to an out-of-date installation, but it needs to go further and provide something more automated. Personally I recommend the Subversion install, for those with command-line access; I used it for the 2.3.1 update and it worked well.

But I digress. The amusing part of Cameron’s post is his link to the comments on a news report describing the defacement. I believe in the value of comments, but some of the leading news sites are afflicted by knee-jerk commenters with time on their hands, who twist every post into another salvo in the OS wars. An news item about a Microsoft “security” expert being hacked seemed an ideal candidate (though I don’t believe identity is the same as security). “This is a shining example why you should host on Linux + Apache,” says one comment.

As Cameron observes, his site and blog is hosted by a third-party and runs on FreeBSD + Apache.

Conclusions? First, the thoughtless commenters on this kind of site are doing the community a disservice, by discouraging others with more interesting contributions.

Second, it shows what some have to put with just because of their association with a particular company.

Third, keep your WordPress patched.

Interview with Matt Mullenweg in the Guardian today. This was done at the Future of Web Apps conference. I enjoyed meeting him. He is open and articulate. I had not appreciated until now that WordPress.com took the opposite decision from Google over the issue of being blocked in countries such as China which are less permissive than the USA about what can be published. He found out that by blocking certain words and tracking certain people the site could be unblocked:

Google had the same decision, and they decided that being there was less evil than not being there, ultimately. For us, we decided that being there under those circumstances isn’t worth it. We’d rather not be there.

A blogging site is not the same as a search engine. It’s arguable that both sites made the right decision. Not easy.

I was also struck by Mullenweg’s espousal of an Apple-like minimalism in software design. He says WordPress has too many options. He was particularly critical of Open Office:

If you open up Open Office, look at the preference screen, there are like 30 or 40 pages of preferences. Stuff that you and I will never care about and should never care about.

I accept the main premise - software should just work. I understand the further implicit argument, that adding options tends to diminish software quality, by adding complexity to the code. But it would be interesting to analyze some of the options in, say, Open Office, and find out why they are there and who is using them. Is having all these options tucked away really a bad thing, or this really more about user interface design?

There are huge numbers of Microsoft bloggers; yet in some important areas Microsoft seems happy to let its opponents make all the noise.

Internet Explorer is an obvious example. There is an official IE Blog, but you won’t find anything there about IE8, just occasional news of minor IE7 tweaks. The comments on the other hand are full of questions, many of them good ones that deserve an answer, or at least an acknowledgement that someone is listening.

I spoke to Microsoft’s Chris Wilson at the Future of Web Apps conference back in February, noting that he gave a “good bridge-building talk”. There have been other similar talks, but little of substance since then. Anyone searching the web for news of browser development and innovation will find little from Microsoft, lots from Mozilla and others.

This is not about Microsoft bashing. Rather, it is about web developers and designers who need to make stuff work. Having some idea about where Microsoft is going with its browser helps with that.

Microsoft needs to rediscover the value of high quality blogging that engages with the community. It is not just IE. Soon after the release of Office 2007 I was among those who reported on performance problems with Outlook. This blog still receives thousands of visits from users who search for why Outlook 2007 is slow. Where were the bloggers from the Outlook team? Months later there was a tech note and patch which helps a little, but Outlook 2007 is still slow and there is no real evidence that the company cares.

What about Open Office XML, viciously attacked by IBM and other sponsors of the rival Open Document Format? Brian Jones has a good marketing blog; yet I’ve seen relatively little technical blogging from the OOXML folk at Microsoft, in response to questions raised.

See also Dave Massy’s blog.

This blog is now running on WordPress 2.3. The differences from 2.2 are minor from the user’s perspective, which strikes me as a sign of maturity: it was already very good. Unfortunately the team did not address my number one wish, which is paged comments - but I realise I am in a minority as the feature has only 8 votes at the time of writing. There is a plug-in, but it does not work well with the theme I use. I am not complaining though; WordPress is fantastic and I am a loyal user. I have started using it for longer articles as well as blog posts; it is effective as a simple content management system for this site.

With the upgrade to 2.3 I have also converted to a Subversion install. This means I can do a Subversion update to grab the latest version, making it easier to stay current.

Matt Mullenwegg is speaking on Wordpress.com architecture at the Future of Web Apps conference in London later this week. I will be there and blogging.