Just added a replies link each comment where there are specific replies. A green background instead of the usual blue background is also now used to highlight specific comments that have been linked to. Hopefully this will make it easier for people to find replies to comments and questions on the site.
I’ve just updated Carsurvey.org with new mobile responsive CSS and HTML5. Desktop users shouldn’t notice many changes, beyond some tweaked fonts and new drop down menus, but mobile and tablet users should now be enjoying a much better experience, which adapts itself to the size of your device. I’ve also taken the opportunity to update all the images on the site so that they render at maximum quality on new high resolution devices like Retina iPhones and iPads. If you have any feedback or encounter any problems with the new layout, please leave a comment on this post or email me (email@example.com).
I know I haven’t posted anything for a long time (sorry about that), but it’s mostly because I’ve been working on my new iPhone weather app Today Is Like, which has just been released today. 15 months in the making, it focuses on comparing today’s temperatures to historical averages from around the world, and in doing so, hopefully adds some interesting context that’s not available in more traditional weather forecasts.
If that sounds interesting, visit the app’s website at http://todayislike.com where’s there’s lots more detail, including a demo video of the app in action.
Just a quick blog post to explain that Carsurvey.org and other CSDO Media sites have moved to a new server this morning. Hopefully the move will go unnoticed by most people, but the underlying Linux distribution has changed (from RedHat Enterprise 5 to Ubuntu 12.04 LTS), so if anyone spots any issues that my testing has missed, please let me know.
I’m not a fan of Google these days. They were the most awesome company in the universe 5 years ago, but their search results have become really crappy (thanks to Panda and Penguin), and their desperate efforts to drag everyone into Google+, remind me of the bad old days of Microsoft trying convert the web to a mess of ActiveX.
But much as I think they’re making some big mistakes, they’re still getting a lot right. I really like my new Nexus 7 tablet, and while I dislike Google’s actual results, their search interface is still awesome.
I’ve been trying to move away from Google over the past 6 months, but each of the alternatives has issues:
Bing: Great results, but the interface is very poor. No easy filtering of recent results is a big problem.
Yahoo: Their homepage is such a mess. And they’re really just a proxy for Bing.
DuckDuckGo: I’ve been using DuckDuckGo as my default Chrome search for the past 6 months. I love the purity of their search, and their bang syntax is awesome, but for finding recent results that are skewed towards the UK, they’re not really competitive with the big boys.
My new solution: I’ve just realised, that if I set my Chrome search to Yahoo! UK & Ireland, I get results from Bing’s index, I get UK targeting and the ability to filter the results by date, and I never have to see Yahoo’s homepage. The results page even looks more like Google’s results than Bing, and now Marissa Mayer is running the show, Yahoo might actually be going places.
Plus, having recently contacted Google, Bing, and Yahoo about various issues, Yahoo were by far the most helpful of the three.
So here in 2012, I can’t quite believe I’m saying this, but the best search experience is Yahoo, powered by Microsoft, through a Google browser!
I’ve just released a very big change to Carsurvey.org.
I’ve spent almost a year and a half trying to escape the effects of Google’s Panda algorithm, with no luck so far.
Having cleaned up the design and ads as much as possible, the only thing left is to try to remove content from the site that may be triggering the Panda algorithm, due to a perceived low quality level. I’ve been doing this manually over the last year, but with so much content on the site, it’s taking too long.
My new solution is to analyse every piece of content on the site with an automated script, which looks for signals that may be indicative of poor quality. Things like unusual sentence lengths, strange punctuation or capitalisation. Content that exhibits enough of these signals has now been removed from the site, pending manual review. In total, about 15,000 reviews and 38,000 comments have been removed. Reviewing and restoring all of that content is likely to take several years.
Sadly, there will be a lot of false positives as a result of this process; perfectly good reviews and comments that just happened to have some characteristics that my script was looking for. If anyone is concerned about the fate of a particular review or comment, please email me (firstname.lastname@example.org), and I’ll prioritise it for restoration to the site.
I’ve just made a massive change to Carsurvey.org. There’s a new theme, that’s HTML5 compliant, with completely fresh CSS (rewritten from scratch, with the help of SASS and Normalize.css). The site has a new logo, and for the first time, has a favicon.
The new theme is designed to be less busy and more consistent, with the content taking priority over ads (there are less ads, and they’re in less prominent positions). The theme is also built around the 980px viewport width assumed by the iPad, so it should work especially well on that device.
There are still some tweaks to be done, but I’m crossing my fingers that visitors will be as enthusiastic about the new theme as I am.
I’ve just debuted a new layout for Carsurvey.org, where instead of each review having its own page, there are multiple reviews per page. This change has been driven by several factors:
In February, Google released an algorithm change known as Panda, which reduces the rank in Google search results of sites that the algorithm judges to be low quality. It’s not exactly clear how Google measures this, but this post from Google is the most detailed description they’ve given:
Carsurvey.org seems to have been penalised by the Panda algorithm, and whilst I believe the site is high quality and useful to visitors, there’s always room for improvement.
Duplicate content due to the site’s layout and quoting in comment replies could have been an issue, and changes have been made to reduce this; mostly by searching for duplicate comments using a longest common substring approach, and replacing large quotes with links that indicate which comment is being replied to.
The site also had lots of pages that only contained a short review or comment, and those weren’t exactly the best experience for visitors. The new layout uses some smart pagination algorithms to make sure that pages that are thin on content will be very rare, and in the exceptional cases where they are unavoidable, they’re marked as NOINDEX so search engines know to ignore them.
Where there are only a few short comments, they’ll be added to the end of review, rather than existing on their own page.
Ads have been reduced in size and number on the site, in case that was causing problems.
Also, the ability to filter comments by whether they are on-topic or not has been removed. It wasn’t being used very much, and didn’t justify the complexity it added to both the user interface, and the underlying code base.
As usual, any feedback would be welcome. This is still a work in progress, and whilst I’m not planning any major changes, there will be refinements coming over the new few months, based on how well the new layout performs with visitors.
It’s that time of year again. Carsurvey.org’s web server has suffered a disk failure. Thankfully all the data seems to be safe. Have disabled new reviews and comments until I’m happy things are back to normal. Expect some downtime soon when the server is turned off to replace the faulty disk.
For years, the emoticons (happy, sad and neutral faces) shown on the lists of reviews on Carsurvey.org have been based purely on the answer to the question “Would you buy another car from this manufacturer?”
Although this question was often a good indication of a reviewer’s overall opinion, there were a significant number of reviews where it didn’t work at all.
In order to address this, the numerical ratings (Performance, Reliability etc) are now also taken into account when deciding which emoticon to display for each review, and this seems to significantly reduce the cases where the emoticons don’t reasonably reflect the general sense of the review.