Slightly ajar

I recently left Opera and now is as good a time as any to have an introspective into the company and its products. Opera does many things right and some things wrong, and here are my personal thoughts now that I'm away from the coalface.

Innovation

Opera has always been an innovative company. In many ways it has often been ahead of its time. Many of the features commonly found in the popular browsers debuted in Opera in one form or other first. Speed Dial is one example, as is being able to open more than one document at once, which morphed into tabs — something most browser users couldn't live without. Even if you look at new features in OS X Lion, some of the key features such as full screen mode and session restore have been in Opera for many years. If you look at recent trends such as webOS (apps made with web technologies), Opera had Opera Platform in 2003. Mozilla are now talking about device APIs, but Opera was very early in that game, has implemented the WAC APIs (which are quite terrible specs) and are fairly heavily involved in the W3C standardisation of the Device APIs via the DAP group. Opera was also on mobile very early, before Apple even invented the whole concept (I guess Opera owns a time machine).

Opera has all this innovation, but beyond mobile and some of the device spaces it has never really taken off as much as it should or deserves to have. There is probably a variety of reasons for this. Some things were just ahead of its time. Opera Platform came out when phones were not powerful enough and browsers were still slow compared to these days with hardware acceleration and JITed JavaScript engines. Opera also doesn't have a hardware partner which makes that kind of project even harder. Opera Unite as another thing that was way ahead of its time and took a lot of resources. I'm not sure the implementation was right either. It seems like as separate apps they almost compete with Opera Widgets and now Extensions for developer mindshare. Making it an API for other products would have been better in my honest opinion.

The desktop browser

It is the desktop browser as a whole that people criticise for not having enough users though. I think many people are missing the point when just looking at market share stats. On a global scale in an industry the size of web browsers, even just 2% of the market is many millions of users. Opera desktop alone has over 50 million active users. That is more than many web sites and services which are lauded for being popular. Opera desktop is also a key player in certain markets such as Russia and the CIS region, where it is competing for the top spot. In Eastern/Central Europe it often hovers around 10% mark, which is nothing to be sniffed at. However, it is true that it lags behind the larger rivals like Chrome and Firefox. This is a fair observation. I think one of the reasons is due to the innovation point raised earlier. Opera has always been good at looking out for the next killer feature and adding innovation to the browser. Innovation is costly however. It is much more expensive to come up with new ideas than to copy someone else's and refine it. Refining the features it has is something Opera has never done enough of, rather than being the quest for new features. Opera desktop probably has enough features now (I'm talking about the browser, not the rendering engine), and it could do with going through each of those and giving them a nip and a tuck, and polishing both how they look, how they work and how they are exposed to the user. I'm quite confident this will happen now.

Opera never used to have a UX group until a couple of years ago. They’re still hiring but they have some good people there already and you can see some of the movement they've made over the last few versions. Opera 11.50 debuted the featherweight UI, which looks like the first phase in cleaning up the UI and making it cleaner and visually lighter. This is only surface deep so far. The main UI now looks fairly polished. The main problems come when you get behind that and start to open side panels, dialogues, and preference windows. These are the areas where they haven't got to optimising yet. Once they've worked their magic right through the app, it will feel like a much more solid product. At the moment it ruins the feeling when you see stretched widgets, messy preferences and the whole issue with the sidebar when the first tab is active. One of Opera's key strengths is it is very customisable, but that makes it even harder for the UX team to design every possible combination so everything looks right. They're well on the way to getting there however.

The other issue beyond the product is marketing. Opera's marketing is often criticised but that is a little unfair. It is mentioned how fast Chrome took off compared to Opera, but it feels that you can't move these days without seeing Chrome ads on every page on the net (only a slight exaggeration). At least here in the UK. Opera can't compete with the kind of exposure that Google can give to Chrome, or MS or Apple can give their browsers by making it the default. Mozilla is the closest you can compare to Opera, but they too used to get advertising from Google before Chrome was released. Mozilla have done a good job at marketing their product though, and appealing to people through the open source nature of the product. I'm not a marketer so I'm not sure what the solution is for Opera, beyond staying creative and spending their limited budget wisely.

Success on mobile

There are no such issues with Opera Mini. It is growing at an incredible rate every month and is the number one mobile browser according to StatCounter. It has been a huge success in the developing world where people often can't afford smartphones and the data is either slow or expensive or both. Opera Mini was designed for exactly these situations, and it has thrived through both word of mouth and operators distributing it to their customers. It saves the users money and the operators bandwidth; a win-win situation. In the top developing economies (BRIC and N-11) Opera Mini is in the top two browsers in every country except China and South Korea (which isn't even a developing country). For all the countries where it is second they have been first within the last 12 months. China is the odd one out, but Opera recently announced a joint venture with Telling, and they are confident they can generate huge numbers of users.

There are only really two dark clouds with Opera Mini. The first is that most of the traction is in the developing world and Opera has never managed to get Opera into the top two browsers in the established developed markets where most of the developer mindshare is located and the ARPU is higher. These are countries where Opera Mobile is more suited. The other dark cloud is that Android is starting to take off in the developing world. Opera Mini faces more and better quality competition on Android than it does on the various feature phone platforms.

Opera needs a strategy to get Opera Mobile more popular in the developed world, and to convert Mini users into Mobile users when they switch to Android in the developing world. The good news is Opera Mobile has Turbo, so users can switch to a browser that has both a familiar UI and has some of the data cost saving attributes of Mini once they switch to a more capable platform that supports fully blown browsers. They have the opportunity to do this but they have to execute before large numbers of users switch to the Android browser. For the moment though the Opera Mini user base is showing no signs of slowing down. Apple have always been good at replacing their own successful product at the hight of its powers with another one (such as killing the iPod Mini and replacing with the Nano or replacing the iPod with the iPhone) before a competitor could to it to them. This is something at Opera could perhaps learn from.

One Web and the engine that runs the vision

The thing that powers Opera and is a core part of their one web vision is the rendering engine: Opera Presto.There has been a lot of external talk about why doesn’t Opera just use WebKit, but Presto is a key asset for Opera and allows them to control their own destiny. They have a lot of amazing developers working on it, and it is competitive with anything else out there, both from a speed, standards support, compliance with those standards and security aspects. Presto is very strong and it will only get better. New features are added regularly and the window between implementation and being included in an actual product has never been shorter.

It used to be the case that a lot of the Core development resources would be taken up doing custom project that sales sold to partners. This would result in a lot of one of code that didn't benefit the rest of the product and took developers away from adding new web standards. That led to a period where Opera fell behind slightly with standards support. This is gladly more or less a thing of the past now, and Presto is right up there with the best or there abouts. Presto has started adding a number of things first again–just like the old days–including the time element, micro data, getUserMedia (formally the device element), CSS Device Adaption, etc.

The Opera Presto rendering engine was built with the aim of running on platforms with limited resources, which has allowed Opera to run places where others have not been able to compete successfully. Its main drawback has often been site compatibility, as many developers do not test in Opera or block it completely. We have made great strides on that, but it still isn't a fully solved problem.

Developer experience

Opera needs help from developers to ensure their sites work in Opera, but Opera also has to help developers. I worked on a number of initiatives to do that, such as Opera Dragonfly, developer evangelism, and Open the Web. We hired a number of web openers in the US recently to try to help developers solve their issues in Opera, and to increase our presence in the US web development scene. There are a number of issues that still make it harder to develop for Opera than it needs to be however. One of the key issues now that Opera has high quality debugging tools is the lack of an open bug tracking system. With any other browser, including IE, you can look at what bugs exist, report your own and track the progress. No so with Opera. There is only a wizard for reporting bugs, but you can not search or track the progress. This lack of feedback stops developers reporting bugs,which leads to more compatibility issues. Exactly the sort of thing Opera doesn't need more of. Opening the bug tracking system is something I tried to get done in the past, but more or less gave up. I wish I had pushed further. Maybe with a new-ish CEO this might be a possibility again. Especially now I'm an outsider and can't access the system anymore myself, I'd love to see it happen. Other issues include some long lasting bugs such as rounding issues and handling of large values but they're starting to be fixed now.

Another area that lacks developer interest is Opera Widgets. They are potentially interesting as a way to develop cross platform apps, but they've never really taken off so far. Part of the issue was the way they were initially implemented, where the browser had to be running first to open a widget. The biggest issues though is I don't think anyone has grabbed the project by the horns and built it into a truly competitive platform. There is still no device APIs in Widgets and the general APIs lack compared to the power of native apps, especially on desktop. They really need access to the hardware, system services, and to have some sort of framework to make building the UI easier for it to be a truly competitive platform. The platform really needs a lot of investment to get to that sort of quality. Some things should come for free though when they're implemented in Presto for other projects such as Opera Desktop and Mobile. Opera Extensions seem to have been a much bigger success, but extensions are often much smaller and easier to make than fully blown applications. The use case is also more obvious as it directly relates to Opera. Opera seems to spread its self too thin sometimes with all the initiatives it has going on, rather than making one solid before moving on to the next. It seems like there is too many different developer platforms going on for a browser with Opera’s market share. If a developer wants to develop for Opera you can either build a regular web page (the Web), an Extension, a Widget, or an Opera Unite service. The use cases don't all overlap, but the amount of choice creates confusion and developer fragmentation. Developers have to think too much before they even get started.

In conclusion

Looking back at everything that has happened at Opera since I started, these are the things I think Opera would benefit from.

I have an incredibly horrible outdated phone. A HTC Touch Diamond to be precise. There is more or less nothing to like about it; especially that I have to pull out the battery to switch off the alarm when it goes off. I’m in the market for a new one, but it is difficult to know what to get. The iPhone 4 looks very nice, but is not much use for me when the manufacturer restrictions mean I can't install my company’s flagship mobile browser. The logical next choice is Android. The platform is more open and its popularity with developers in the US mean that there are plenty of apps for it. Unfortunately there are no Android phones with as nice industrial design (in my opinion) as the iPhone 4, the top of the range HTCs are getting a little long in the tooth now, and the latest Motorola Droids are not out in Europe as far as I can tell.I guess there will be a device (probably from HTC or Samsung) which will match then beat the iPhone’s DPI in the near future, especially as they left the iPhone 3GS’ display looking ancient for quite a while by the time the iPhone 4 came out. That has not come yet though.

Looking at so many phones has made me think what would be my ideal phone, especially if money was no object (which sadly it is). Speaking of money being no object, Engadget had a post today on the 1 million dollar phone. The design is as vulgar as the price tag. Not only that, but like other super-premium priced mobiles, it packs everything around what is basically a low-end heap of junk feature phone. You’d think if people were willing to pay tens, or hundreds of thousands of dollars, they'd at least want a phone that wasn't outdated 4 years before they bought it.

But this did phone did get me thinking. The value is all in the materials like a piece of jewellery, but being a piece of electronics it will be totally obsolete at some stage, and it seems a waste to throw away hundreds of thousands of dollars of black diamonds away. Why not instead invent a phone which was designed to be long lasting and factory upgradable. Just like how you send in your watch to be serviced or how high end Leica M-System cameras can be sent back the factory for upgrades, imagine if we could do this with phones?

I’d love to be able to buy a phone which has a simple design aesthetic, such as Dieter Rams’ work for Braun or Kenya Hara’s work. A cross between Scandinavian minimalism, the Japanese Wabi-sabi and Wabi-cha design aesthetics, and German engineering. Materials are the key here, along with clean lines which don't date quickly with passing fads. I think Apple are still one of the very few hardware companies that understand how important good materials are. OFten companies ruin promising designs with cheap looking plastic. Use materials which are long lasting and natural where possible. Metals, wood veneers, leather or ceramics could look great if done well, with restraint. The trick to my ideal phone would be the internals. Pay a bit of a premium for the externals for something durable and long lasting with a high quality feel, and include components inside that can be upgraded in life-cycles rather than the full phone being thrown away. We’re at the stage now where the screen is more or less the size of the face of the phone and it is just the pixel density going up in a similar form factor, unless you want a bigger phone. We’re also at the stage where phones are not getting too much thinner. There is only so thin a phone can get before it becomes uncomfortable to hold. Obviously there is still a lot of room for slider phones to get thinner as th keyboards still add a lt of heft. If you could just send the phone back when you decide to upgrade, and get say a faster processor, more memory (probably combined with a new circuit board for those upgrades) a higher res screen and an improved camera sensor, have the case polished and cleaned and the warranty extended by a year, that would be fantastic. Less waste and packaging too, especially if there was some way to repurpose the replaced parts. You'd only have to replace the whole phone when you are sick of the design or the form factor becomes out of date. There could perhaps even be ways to adapt the form factor of the case if well designed. The issue would be labour costs with upgrading (so probably would have to have a decent premium) and it may have to be more bulky if parts were replaceable rather than soldered onto the motherboard or such. It would also perhaps be hell for developers too if any component could be upgraded at any time.

If something like this could work then you could keep a phone for years, just like a good watch, and be very personaliseable to the needs of the user. I know I'd buy one if I could afford one and had great industrial design. As long as it doesn't use Windows Monbile 6.2…

While messing around with SVG I created a map of Opera market share in Eastern and Central Europe. This is much heavier than the previous SVG files I've made as I started with an existing SVG map from Wikipedia, that was originally produced using Inkscape. This adds a lot of extra things that machine generated markup usually produces, but making a map by hand would be bordering on insanity. I've cleaned up some of the markup, but I should be able to trim the file size a lot more by removing some of the inline and redundant styles and extra elements and attributes that are not needed.

I added all the extra information, such as the labels, colouring and animations by hand, which isn't too difficult to do once I worked out that the map was using a viewBox that changes the co-ordinate system (everything I was trying to include was too small to be seen because of this). I'm using the alpha channel of the HSLA colour model to show the different levels of marketshare (offset by 0.2 to make the colours easier to spot), which gave a easy 100% scale to work with. This is easy to do, but have performance penalties and opacity is expensive. It also means that it requires Opera 10 or greater to get the full effect. I'll probably end up mapping the colours to RGB values and use that instead, although that would make it more expensive. The colouring scheme chosen also makes it slightly difficult to see the difference between the colours less than 10% and those around 10 to 20%, so i reduced the opacity of the country borders for those less than 10%. I'd eventually like to show the market share data when clicking on the country rather than a predefined animation that loops through each county. In theory you can zoom in on each country (for those I added - Czech Republic, Slovakia, and Poland), due to the use of the view element and fragment identifiers on the a element, but support for this seems to break when animations are included. I'm not sure if this is a bug or something I'm doing wrong. You can still directly go to the county using a direct link outside of the document. There is also an issue with Safari where HSLA colour model doesn't seem to be supported in SVG, or at least not on the fill property, so the map just displays as black.

As far as the data is concerned, I took the market share values from the StatCounter results for April (at the time of writing). All figures are rounded so 2.49% becomes 2% and 2.51% would become 3%. You'll notice some patterns on the map, with 3 main groups. Those countries that are in the CIS are all post 20% market share. The Visgrád_Group and Baltic nations are around 10% to 15%. Latvia is slightly lower with 9%, and Estonia drops off at the edge of the group with 7%. The major exception is Hungary with only 4%, but we have identified a number of compatibility issues in Hungary (see comments in my previous pot) which may be causing this. The third group is the Balkan and other South-Eastern countries, that are around 4, 5 or 6% market share. This again has exceptions with Serbia and Montenegro both being ahead with around 10% market share, and Slovenia and Albania around 2% market share. I wouldn't be surprised if by the end of the year Hungary has caught up to the 10% mark and the third group of countries also start working their way to that level.

[UPDATE 1] HSL and HSLA are apparently not suppose to be supported by SVG, although they are both supported by Opera and Firefox. I've switched the fill value to red and added the appropriate alpha channel value with the appropriate fill-opacity value. Next up is the issue with Firefox. after talking to Jeff Schiller, its been narrowed down to the lines used to draw the grid. Firefox treats these differently, but I have to work out why. That is something for another day though.

The Baltic states are made up of Estonia, Latvia and Lithuania in the UN region of Northern Europe (along with the Nordic nations and the British Isles). These three nations are unique in that they are the only former Soviet republics who are not part of the modern day CIS. They are also the only former Soviet states that are members of the European Union.

When looking at Opera market share in the Baltics, we have two sources we can compare, StatCounter and Gemius, which I also included for Ukraine in the CIS edition of this series. With two sources, if the results are comparable, I can be more confident of the accuracy than the previous post on the former Yugoslavia. As Gemius focuses on Eastern and Central Europe, they should have a decent sample rate for these countries. Lets see how Opera fares in the third region in this study.

Lithuania

The largest nation in the Baltic region by Internet population is Lithuania, with 1,333,000 Internet users in 2007 according to the CIA. If we take a look at StatCounter figures, Opera had 11.92% market share in Lithuania in February. As is common with the results in the previous study, this puts Opera in third behind IE and Firefox and more market share than Safari and Chrome combined. Over at Gemius, Opera had 7.9% market share in February, and is also in third place ahead of Chrome and Safari.

Latvia

Latvia is closely behind Lithuania with 1,177,000 Internet users in 2007 according to the CIA. By StatCounter figures, Opera had 8.81% of the Latvian market share in February. As with Lithuania, Opera was in third place overall, this time with Firefox in first place. Opera again was bigger than Safari and Chrome combined. The Gemius figures show that Opera had a 4.8% market share in Latvia in February. Opera was also in third place by Gemius figures.

Estonia

With 780,000 Internet users in 2007 by CIA stats, Estonia is the smallest of the Batlic states. If we look at StatCounter figures we will see a common trend from the other two nations, with Opera in third place behind IE and Firefox and more than Safari and Chrome combined, with 6.02% market share in February. Opera doesn’t fare so well in Gemius figures with only 2.4% market share in February, but again it is in third position.

Overall

While Opera doesn't reach the heights of its CIS market share or rankings, Opera has a solid following in the Baltics, with similar figures to the former Yugoslavia. Opera is again constantly bigger than Safari and Chrome combined, and is constantly growing. By StatCounter figures, Opera already has double digit market share in Lithuania, and is pushing towards the 10% mark in Latvia. Estonia is the weakest of the group, but Estonian localisation of Opera only came in Opera 9.6. It will be interesting to see if this will make a difference throughout 2009.

Often web developers will say something like Opera is a great browser, but we don’t care about it, as it has less than 1% market share. The stat that is often pointed to is the one from Net Applications. Net Applications of course are primarily US and English language focused (indeed their web site seems to only be available in English), and in the past they have retroactively reduced Opera’s market share in their statistics. Market share stats should always be taken with a grain of salt, as there are many variables to take into account, from sample size, geographical bias, detection techniques, spoofing, et criteria.

There is now a new player in town in the market share stats game, StatCounter, the popular visitor stats tool. In this post I will look at how StatsCounter and other free market share sites rank Opera in the CIS, which is one of Opera’s strongest markets, and a big chunk of what was the Soviet Union. I’ve started with this region, as it is a strong market for Opera, but in further posts I’ll look into further markets. I will however state up front that StatCounter ranks Opera as third browser overall globally ahead of both Safari and Chrome on almost 3% market share. I believe that Stat Counter is still more anglophone centric, so that figure could be a little low, but Stat Counter are working hard on localisation, with the Hungarian version already released, so I expect the bias in favour of US market share to even out over time.

There are three sites I’m using for this comparison, the aforementioned StatCounter, Gemius, which has free stats for a few Eastern European countries, and LiveInternet.ru which is a popular Russian language based stats site. Not all of these sites cover all markets unfortunately, but using multiple sources gives a fairer picture. I’ve not included Net Applications, as the free stats don’t include country by country breakdown. It is the same case for XiTi, which was last updated in November, but ranks Opera fairly strongly as 5.1% for the countries in Europe it monitors.

Russia

The biggest market in the CIS is of course Russia. It has one of the fastest growing internet population in the world. It had a Internet population of around 30 million in 2007, according to the CIA World Factbook. So how is Opera doing in Russia? According to StatCounter, Opera has almost 38% market share in Russia, in February, making it the number 1 browser ahead of Internet Explorer. LiveInternet on the other hand ranks Opera desktop as 21.5% market share, second behind IE. Gemius doesn’t track Russia. Overall Opera has a strong showing in Russia, placed first and second respectively. You’ll also notice the strong showing of Opera Mini in the LiveInternet stats.

Ukraine

Ukraine is perhaps the second biggest market in the CIS, with a Internet population of 10 million in 2007. Opera is doing even better in Ukraine than in Russia. By StatCounter stats, Opera has almost 44% market share in Ukraine, in February, again making it the most used browser ahead of IE. According to LiveInternet, Opera has almost 30% market share in February, second placed overall behind IE, but Opera 9 is the most used browser version, ahead of IE7. Ukraine is the only country in the CIS that Gemius covers. On this site Opera has 29% market share in February, second behind IE, but again Opera 9 is the most used browser version in Ukraine. Again, overall, Opera is very strong in Ukraine.

Belarus

Belarus is the other major CIS state in Europe, with a Internet population of 6 million in 2007. Again, Belarus is a very strong market for Opera. According to StatCounter, Opera has just over 50% market share in Belarus in February, making it again the top browser in the country. With LiveInternet stats Opera has 36.4% market share in Belarus in February, again putting Opera in second place, and again Opera 9 is the most used browser version in the country. Opera is certainly very popular in Belarus.

Moldova

Moldova is the last of the CIS nations in Europe, with a Internet population of 700,000 in 2007. StatCounter is the only site that tracks Moldova, and unfortunately Opera doesn’t come out on top, with only 23% market share in February. Opera was beaten off the top spot for the first time, slipping to third behind IE and Firefox.

Kazakhstan

Kazakhstan is perhaps the biggest market in Asian CIS with 1,901,000 internet users in 2006 (if you count Russia as being European). By StatCounter statistics, Opera has just over 39% in Kazakhstan in February, making it the most used browser for the 4^th time, ahead of IE yet again. LiveInternet follows the pattern again, making Opera the second most used desktop browser in Kazakhstan with 23.2% market share in February, but this time Opera 9 is third behind both IE 7 and IE 6.

Uzbekistan

By CIA numbers Uzbekistan has a larger Internet population than Kazakhstan with 2,100,000 connected to the Internet, but the figure is from 2008 instead of 2006. Opera again does well in Uzbekistan by StatCounter numbers with just over 43% market share in February, beating out IE to first place for the 5^th time in this post.

Azerbaijan

Azerbaijan sneaks in over the million mark with 1,036,000 Internet users according to the CIA in 2007. According to StatCounter, Opera fails for the second time to reach pole position, behind IE in second place with 29.14% market share in Feburary.

Kyrgyzstan

Kyrgyzstan has 750,000 Internet users in 2007 according to the CIA. Opera again finishes in second place by StatCounter stats, with 24.5% market share in February, again behind IE.

Georgia

Georgia is perhaps most famous recently for being the centre of news coverage over the last year or so. Georgia had 360,000 Internet users in 2007. Opera again finishes in second place behind IE with 39.37% market share in February. Georgia had large growth in Opera users last year, so I expect that the 360,000 Internet population is well out of date now - I suspect it is much higher.

Armenia

Armenia is a relatively small country with 172,800 Internet users in 2006. Like Georgia, this figure is likely well out of date now. Like Moldova, Opera slips to 3^rd for the second time, again behind IE in first and Firefox in second in statCounter statistics. By StatCounter figures Opera had 27.36% market share in February, closing in on Firefox for second position.

Turkmenistan

Turkmenistan has a small Internet population of 70,000 in 2007, making it the second smallest of the former Soviet republics. Opera finishes second in StatCounter stats, behind IE with 28.43% market share in Feburary.

Tajikistan

Tajikistan is the smallest of the CIS nations with 19,500 Internet users in 2005, but due to the age of this statistic, it is likely larger now. In Tajikistan StatCounter ranks Opera as second behind IE with 34.76% market share in February, rounding off a successful region for Opera in terms of users and market share.

Overall

It is clear to see, using multiple sources where possible, that Opera is very successful in the CIS region. It regularly ranks first or second in terms of desktop browser market share in StatCounter stats and first or second in terms of browser version in LiveInternet stats. LiveInternet stats are probably more accurate at present as they focus on the Russian Language market, and StatCounter probably has a smaller sample size for each of the countries in this study. As StatCounter localises, I’d expect the sample sizes to go up, which may lower the Opera market share slightly, but it will make the CIS and other none US samples to be more representative and thus push up Opera’s global market share. Another reason why StatCounter’s stats are higher than LiveInternet may be that StatCounter could include Opera Mini in the Opera stats. If this is the case then the two sets of figures would be closer together.