Easyjet has recently updated its booking site to use a newer, slicker AJAX/Web 2.0 GUI

Unfortunately the new booking process seems to be more hassle and slower than the old one. More than once when I used it recently I had to hit reload to get a page to display or to get the next page to show up.

I’m not singling out Easyjet for criticism here, I’ve seen similar “improvements” on other sites, both eCommerce and general interest. Part of the problem is that web 2.0 seems to load web servers differently (not necessarily more but placing the load in different places) and so when a site moves to the new interface various services are hit harder than they used to be and get overloaded. In particular web 2.0 sites seem to make more, but smaller, database transactions and I suspect that this is the heart of the problem as the existing database will be tuned to larger but fewer transactions.

To add to this it is entirely possible that Web 2.0 sites get hit by a perception in slowness that is in fact due to their attempts to provide a faster service. Because a Web 2.0 page will load (as far as the browser is concerned) quicker the delay as the bits of the middle of the page are loaded on top becomes noticeable. When there was just one loading thing at the top even if it was as slow or slower we put it down to network issues as opposed to something on the page, thus we were more tolerant of the slowness.

Of course sometimes it is the lack of horsepower in the client that causes the slowdown. By moving more of the processing to the client, web 2.0 sites suffer when clients lack the speed and resources needed to display the page. Some sites with masses of Web 2.0 content such as facebook and gmail have come up with special “lite” versions which load faster and do not place such a burden on the client browser. We may well see more of this as time goes on.

One of the undoubted benefits of Google’s entry into the web-browser world with Chrome is that it has helped light a fire under the other broswer providers. Google’s reason for entering the browser wars was that it needed a better javascript engine to run the sorts of javascript heavy pages that its online applications use since javascript performance was frequently less than stellar.

This has had an effect with most major browser providers increasing the speed of their offering (interestingly while earlier tests had Chrome fastest, in a recent benchmark Apple’s Safari betas seemed to be slightly faster that Chrome) but it isn’t the only way that Google has influenced the browser wars. The other thing Google has done is given a kick in the pants to a number of HTML 5 features that might otherwise have languished.

Interestingly one of those features - downloadable fonts - is not yet supported by Chrome (though it is supported by Opera’s v10 beta and Firefox 3.5 (and supported differently by IE 8)). A couple of other features - the video and audio tags - are only available in the Chrome 3.0 alphas. While downloadable fonts are probably not that exciting - though this page explains why designers should cheer - the video and audio tags clearly have a major potential hit on Adobe because it pretty much removes one of the main drivers for flash. This recent Cringely column points out a related attack on flash via the codecs used and should probably be read as well.

Indeed the combination of downloadable fonts, javascript to download XML etc. and these tags mean that flash becomes almost entirely superfluous for web site creators. If web viewers don’t need flash for video or audio and web creators utilize HTML 5 to make flash like page effects natively in the browser then flash loses its status as a default plugin that all browsers have. Amusingly this may end up helping Microsoft because it potentially allows Silverlight to compete on a more level platform. On the other hand it is extremely unclear to me what benefits Silverlight will bring to the table that the same combination of natively supported browser features cannot also deliver.

The problem here, for both Adobe and Microsoft is that their proprietary development environments now look less like a basic requirement for a full service deb design house and more like an optional extra. It will be truly fascinating to see whether or not sales of Flash and Silverlight development tools drop.

On the other hand Adobe’s AIR potentially allows non browser based applications and applets to use the same AJAX and flash technologies that are used in web pages. The question then becomes whether users prefer to have 101 browser tabs/windows or 101 separate applets?

The current (online) edition of IEEE Spectrum magazine reports a very interesting “breakthrough” in memory technology. It is reported that a team of physicists the University of California, Berkeley, has developed an electromechanical memory device based on “carbon nanotubes”. Its developers, led by Alex Zettl, claim the reliable storage of data for up to a BILLION years!

Memory states are represented by the physical position of an iron “nanocrystal” inside a hollow carbon “nanotube”. If the iron crystal is at one end of the tube then a “1″ is represented. If it is at the other then we have a “0″ state. It is said that an expected data density of 1 terabyte per square inch can be expected in practical devices!

To me this seems a major breakthrough. Long-term storage of digital data is a serious problem for the whole of society. Most records today are created in digital form. In addition, all around the world people are digitising old (paper) records and making them available on the Internet. The problem is that we really don’t have a good technology for storing digital information reliably for any reasonably long period of time.

• Data stored on any magnetic medium (tape or disk) does decay over time. Personal experience shows that “Floppy Disks” recorded in the 1990’s have already decayed to the point where they cannot be read. Some people claim that magnetic tapes might last for 100 years - past experience suggests that this is more like 20 years. Albeit if someone came up with an archival magnetic tape storage technology using a serious amount of “Forward Error Correction” then perhaps we might think about 100 years.
• CDs can be very good. But plastic decays and deforms over time and the reflective backing (critical to the operation of the CD) is usually made of aluminium. If air can get to the aluminium it decays and you get the phenomenon called “CD Rot”. This has been a problem for CDs already. However, the CD architecture contains a very high level of “Forward Error Correction” (FEC). Originally intended to overcome the problem of dust and scratches on “music CDs”, the FEC technology used is brilliant! Kodak claims a 300 year life for its archival CD’s with a “gold” reflective layer (which will not “rot”). Of course then there is the potential longevity (or lack thereof) of the dyes used in “recordable” CDs. Of course to get any kind of longevity from a CD you have to store it in a dark place with stable temperature.
• One of the ways that DVDs got their higher capacity (versus CD) was to remove some of the “Forward Error Correction”. Thus Kodak only claims 100 years for its recordable DVDs.
• Flash memories will also decay over time.

Many people make the (legitimate point) that in years to come it is very unlikely that hardware will be available that can read today’s media or that there will be software that understands it. A good point! They say that all digital media should be refreshed (by copying and perhaps re-formatting) at 10 or 20 year intervals. Most of us do this with our business information simply because we change computers every 5 years and have to migrate the data. But how many businesses just store old tapes and disks expecting that the data will be recoverable if/when the time arrives?

Nevertheless - problems of reading the media and understanding the data formats are problems for the people of the future. Today’s problem is to save the information.

With “Carbon Nanotube Storage” it seems that we finally have a way.

Science Fiction writers are, as a general rule, supposed to have a handle on the future and often this is indeed so. If they come from a computer/high-tech background then this is even more the case. Hence the recent speech by Charlie Stross at LOGIN 2009 is well worth reading and pondering.

The speech is about the future of computing, and particularly gaming, over the next 20 years and it has a number of somewhat controversial predictions. One of which is that Mr Stross predicts that we are approaching the end of Moore’s law with regard to chip power and density. Another, perhaps less controversial point, is that the race for CPU manufacturers and architectures appears to now be at “Game over” with two winners: Intel and ARM. What Mr Stross points out that is frequently missed though is that if one counts raw numbers then ARM is in fact the out and out winner. This is because ARM utterly dominates the modern embedded CPU market and embedded devices are far more numerous than PCs, laptops and servers. ARM’s numerical domination has actually been true for (almost) all of the last decade although in the beginning the major CPU types were venerable 8-bit designs. These days new designs all use ARM cores primarily for power reasons as it is rare than a controller actually needs the memory and CPU power of an 32 bit core.

More interestingly he points out that while processing power may be plateauing network bandwidth still shows plenty of potential for multiple generations of growth. Current bandwidth is in the Gigabit/second range and physics only starts predicting problems when we get in the multiple terabits/second range. Of course some media are probably almost maxed out (copper wiring for example) but wireless and fiber transmissions still have a long way to go before hitting limits.

I won’t descibe much more, really it is better to read the original and the comment thread it inspired.

There is a recent article in the NY Times about the threat to the movie industry of “digital piracy”. As is all too typical of modern journalism it argues from anecdote rather than data, and rather more unfortunately, as Mike Masnick from Techdirt noticed, its lead anecdote is spectacularly misleading. The NYT piece starts by noting that the anit-priacy efforts for last year’s movie Dark Knight were a miserable failure:

The campaign failed miserably. By the end of the year, illegal copies of the Batman movie had been downloaded more than seven million times around the world, according to the media measurement firm BigChampagne, turning it into a visible symbol of Hollywood’s helplessness against the growing problem of online video piracy.

It is no doubt true that Dark Knight was widely downloaded over the internet for free. However what the NY Times fails to mention is that Dark Knight was the largest grossing movie of 2008 and the second largest domestic (US) ever behind Titanic.

Given this commercial success could it not be the case that the downloads actually helped sales? In other situations this kind of downloading popularity is something that marketing executives and publiscists dream of because it is the heart of their “viral marketing” campaigns. The article goes on to note that many studios are now deliberately making shows available on line and mentions YouTube and other video sharing sites and notes that:

[M]edia companies are [...] losing the battle over illicit copies of full-length TV episodes and films. The Motion Picture Association of America says that illegal downloads and streams are now responsible for about 40 percent of the revenue the industry loses annually as a result of piracy.

This claim seems somewhat disingenuous. The revenue “lost” is a nominal sum arrived at by calculating what revenue might be obtained if every download were a $10 movie ticketor $15 DVD and ignores the obvious fact that people simply would not watch content if they had to pay $10 for it. Indeed giving stuff away for free (the time tested technique of the free sample used since the days of 1001 Arabian nights) is a highly regarded marketing technique and one that works remarkably well in the digital age. Monty Python recently created a dedicated YouTube channel where official copies of their output can be seen by one and all for free. This move caused considerable buzz (i.e. viral marketing) and the result is that their DVD sales increased by 23,000% at Amazon.

The NY Times says that some prefer to call movie/music piracy “digital theft”, pardon me if I suggest that actually it might be better known as “Unauthorized Viral Marketing”.

The problem with Web 2.0 and “cloud computing” has always been that sometimes you just don’t have internet connectivity. Hitherto most previous attempts at server-centric computing have failed. Enough critical users are on the road enough of the time that server only solutions have failed - remember Oracle’s doomed efforts or a decade ago? The same potential problem has hit web 2.0 applications as well. Salesforce.com has a dedicated offline mode and synchronization method for example and many other web 2.0 applications simply don’t work offline. This obviously limits their use in an increasingly mobile world.

However now the price of serious amounts of memory and storage (flash drives at a few Euros/dollars/swiss francs/pounds per gigabyte and gigabytes of DRAM similarly priced) means that we ought to be able to cache “most recently used” stuff locally and rely on the cloud as a repository of the rest. Until recently all this was irrelevant because the only platform we could use was the laptop with its large hard disk, Windows OS, Microsoft Office suite and comparatively limited battery life. Given the full service Windows OS and Office suite there really wasn’t much incentive in Web 2.0 / cloud versions of basic office productivity software.

Now, however, we have the netbook. Netbooks are frequently linux based and hence don’t have Microsoft Outlook/ Word etc. They also have less memory/storage compared to today’s laptops (perhaps about the same as was common 3-5 years ago) and most important of all they cost a LOT less. They also weigh a lot less and are thus ideal for business trips. Their real drawback though is that they are underpowered and so when back in the office the user wants his real computer - and thus faces the “Sync” challenge.

This combination means that ideal solution is some kind of cloud computing with offline storage. And Google is now able to offer this for Gmail accounts. Not only that but it offers “flaky” mode support for travellers with intermittent internet access. As someone who was recently in that situation I can say that this looks like a real winner.

It may also be the bullet that kills Microsoft. But that is the subject of a separate post

Up until about a year ago the received wisdom of the last couple of decades indicated that the twin powerhouses of personal computing - Microsoft and Intel - would be able to keep pushing the hardware performance/software complexity equation up and up. This factor would keep them and everyone else in their value chains happy because we’d all have to keep on buying new computers to support the new software etc. Witness MS Vista which requires multiple gigabytes of memory and multiple CPU cores to run well.

But about a year ago it all started to go wrong. The first thing was the mis-selling of “Vista Compatible” hardware which many people bought and then discovered could only run Vista extremely slowly.  Of course if you added the RAM then things got better but that required people paying more for their hardware and so they often looked for alternatives. One alternative was to continue to run Windows XP and that put a distinct crimp in Microsoft’s plans. But that may have been the least painful option.

The more painful one was for people to switch to Linux. I’ve been running a flavor of Ubuntu linux for close to a year now on machines that I use heavily. It performs well and I can do almost everything I need to do without any trouble. For the few things that really require a Microsoft OS there is VMWare player running a virtual XP machine. This actually has significant benefits because it means I’ll never need to fiddle with reinstallation hassles, “Windows Genuine Advantage” or similar ever again even when I upgrade from one PC to another.

All of this hurts Microsoft but may not necessarily hurt Intel. After all even with linux there are applications (e.g. photo editing) and scenarios (virtual machines) where a lot of CPU horsepower still makes sense.

Which leads us on to the Netbook. The Netbook niche - popularized by the Asus eee - runs linux on low spec hardware and makes no attempt to support applications and scenarios where massive CPU resources are required. The netbook products have certainly helped spread linux - and also helped keep Windows XP alive - but they may now also be beginning to hurt hardware too. Intel’s Atom chipsets which are used in newer Netbooks have been phenomenally successful but they could be hurting Intel’s sales of higher powered higher margin CPUs in the process.

Even worse for both Intel and Microsoft is that a machine with an ARM CPU core and Linux/Unix OS could turn out to be even more popular in these economically straitened times. This isn’t far-fetched, as the iPhone is precisely that and is entirely powerful enough to display video, surf the web etc. as are te Nokia N series tablets (both are considerably more powerful and capable than Wintel PCs from 10 years ago but run on a fraction of the power)

The threat here is that just as the Wintel duo kicked minicomputers and unix workstations into an ever shrnking upper niche so now they could see the same thing happen to them thanks to a combination of Linux and ARM.

I’ve had a Bookeen Cybook3 for about 9 months now and read a lot of words on it. The Cybook has an e-Ink display (same as the Kindle and Sony readers) as opposed to the LCD displays of cellphones, PDAs and laptops and this is, in my opinion, the great differentiator between these readers and other portable reading devices.

E-Ink is a static display technology that consumes no power when the screen is not changing. This means that the battery life for a e-Ink reader is in the days (if not weeks) rather than hours of an LCD screen been used all the time. I find that when reading a lot (e.g. when travelling) I tend to need to recharge the reader after about 4-5 days of use. At times when I’m reading less it goes weeks without a recharge. Another e-Ink difference: since e-Ink is a reflective screen it is perfectly legible in sunshine and bright lights, just like paper, and completely unlike LCD screens as this photo illustrates

One issue I’ve discovered in the last month is that the e-Ink display seems to be slightly more sluggish in cold temperatures. I used it while waiting for trains in Zurich and the UK when temperatures were around freezing and page turns seemed to be slightly slower with occasional shadows of the previous image visible. The effect was not enough to impair the reading experience but it worth noting as I assume this would get worse if I were waiting in places like Chicago or Helsinki where the temperatures are likely to be well below freezing. It also seems to me that the Cybook battery, as with most batteries, was less capable in these cold conditions meaning that my time between recharge dropped from 4 days to 3.

While my stints of cold commuting did show up these slight weaknesses they also, absolutely, highlighted its strengths. The Cybook in its leather cover is about the size of a thin trade paperback book and thus fits well into overcoat pockets. It became natural for me to read while waiting for the train, pop it in the pocket while I boarded and then pull it out to continue reading once I had found a seat. Having it in a pocket also meant it was always available for those other periods when I had to wait - for example while my father was being examined at the hospital.

The Cybook has options to auto power off and I have it set to do so after 15 minutes. Once powered off it takes about 30 seconds to boot up. Although this boot time is short compared to a laptop it is noticeable and I may decide to disable the auto power off for a bit and see what that does to the battery life. With the 30 second boot I find that I tend to switch it on in anticipation that I’m going to have time to read in the next few minutes and this seems to work well even though sometimes of course I don’t quite get around to reading from it in time. The boot time seems to be dependent on the number of books in the reader. When I had no more than a couple of dozen books boot time was around 20 seconds with some 10 seconds being added now that I have over 150 books.

The Cybook has a SD slot which takes a maximum of a 2GB SD card. My 150+ books uses about 7% of that space, so it looks like I could have something like 2000 books on the device if I really wanted to. In practice I’m not going to have anything like that amount as, in addition to the boot time, navigation in the library makes that number of books hard to handle. The library will only display 20 books at a time and there is no quick way to go to a particular library page. This means that if you were reading a book at the end of the library list and now want to read another near the start it can take a while to page through.

The other question that arises when it comes to eBooks is whether content is available in a suitable form - and at a reasonable price. I find that the Cybook’s support of plain text, PDF, HTML and Mobipocket means that almost everything I want to read is easy to obtain - mostly in DRM-free formats. The big issue though is price. Many publishers seem to think that eBooks should retail for a price similar to the hardback price (i.e. over $US20) which is ridiculous. In some cases it is cheaper for me to buy the book in hardcover and pay amazon to ship it than it is to buy the electronic copy which is completely stupid. This does, however, seem to be changing gradually.

Overall I’m happy with the hardware, satisfied with the software and happy with the content I can get for it.

It first occurred to me while I was reading stories at the Inquirer that some of the time when I was reading the story the fan on my laptop switched on. Odd, I thought, but not much more. Then I noticed the same thing happening on other sites. Not all, and even on sites where it happens not every page. Odd, I thought, but not much more. Finally the light went on. Pages with some sorts of Adobe Flash adverts are the culprits. When you visit a page with the right (wrong?) sort of flash, CPU usage goes up enormously even though you are doing nothing more strenuous that occasionally pressing the down arrow. In the extended entry you can see my CPU jump from an average of under 10% to about 60% at a particular moment. You’ll have to trust me on this but that moment is when the page with the flash ad rendered itself the first time.

I can understand a peak when the page loads. What I don’t have any sympathy for is seeing it remain high. There seems to be no reason for it other than (I guess) sloppy programming somewhere. I should note it isn’t all flash pages, or even all flash ads, but it seems to be consistent with ads from particular vendors. Lenovo is one, but it is certainly not the only one.

The moral of this story? If there is one it is perhaps that flash causes global warming …

The Register and Bruce Schneier both comment on a recent announcement from the EPFL in Lausanne that a joint team from the EPFL, the University of Bonn and NTT in Japan have managed to factor a 307 digit (decimal) number - a number about the size of the numbers used in 1024 bit RSA encryption. The number factored was a “special” one in that it was in the form of 2ⁿ-1, so it is not one that would actually be used for encryption. However, as the lead researcher explains, while this is not a sign that anyone can factor an arbitrary 1024 bit number today, it is most definitely a sign that the writing is on the wall for 1024 bit keys, because the progress from factoring special numbers using specially gather CPU resources to factoring arbitrary ones of a similar size using more easily obtainable CPU clusters is well known. It seems likely that by 2010 the 1024 bit RSA key will be considered insecure.