12 May 2009 - 15:44I'm currently engaged in iPhone development which I might blog about here from time to time. I know there isn't exactly a shortage of people blogging about the iPhone right now, but still, if I come up with anything interesting I figure it's a good idea to share :)
Comment spam: wake up and smell the Hashcash coffee
22 May 2007 - 18:29
This blog (and other's on exaflop.org) use weblog software called Pivot. A number of versions back, Pivot implemented a system called hashcash to defeat the comment spambots that are the scourge of the bloggosphere.
Lets compare hashcash to the alternatives:-
- Keyword blocking: this as ineffective in blogging as it was in the email world before it. It'll catch many spams, but it will also let many through. The blog owner has to keep looking through the spams and adding more and more keywords to the block list. Eventually you have to stop adding keywords or nobody will be able to add anything to the blog!
- IP blocking: this is ineffective as well because spamming is typically performed by zombie botnets (arrays of PCs that are infected with malware that follow the instructions of remote users while appearing to their owners to work normally). The spams appear from all manner of different IP addresses and besides, you still need to delete all the spams by hand with this method.
- CAPTCHAs - Completely Automated Public Turing Test to tell Computers and Humans Apart. These show the commenter an image that contains a string of letters and numbers. The user has to type this string input a field on the comment form along with their other details. There are two problems with CAPTCHAs. The first is that they are a usability nightmare - nobody likes having to pass a test like this every time they submit a form and for people with sight problems it can be impossible to pass. The second problem is that OCR (optical character recognition) techniques can used to defeat the test. As a result of 2, CAPTCHAs have been made progressively more and more difficult to pass, exacerbating problem 1.
- Bayesian filtering: This is the same as the most popular method of email filtering. A mathematical analysis is made to try and recognise if the comment looks like a spam in the same sort of way that we would recognise many spams as spam without even reading them - we recognise many different signs such as strings of garbage characters, implausible names and other features. This works about as well as it does with email spam, i.e. it produces some false positives (real comments tagged as spam) and false negatives (spams that get through).
As you can see from the above list, two of the four main alternative approaches are completely useless and the other two have serious problems of usability and effectiveness. Hashcash by comparison has been completely effective. That needs more emphasis really. In a year I have had less than 10 spam comments appear on my blog. That is few enough for me to believe that the spams that did get through were entered manually. That is something I can live with.
Akismet is a Bayesian filter system used by many bloggers including those on Wordpress. It uses a 'hive mind' approach that combines spam data from many users to improve effectiveness. Even so, lately I am seeing several bloggers (most notably Robert Scoble) complaining about Akismet either not filtering out all the spam, or catching too many genuine comments in it's filter. Apparently there is (or at least has been in the past) a Hashcash plugin for Wordpress. I would strongly suggest people check out this option. Akismet is a nice idea, but it is clearly not as effective as it should be. Hashcash is effective. I don't doubt that Scoble gets more attempts on his blog than I do on mine, but the results should scale because 100% effective scales.
The final issue then is that spammers will someday build spambots that can defeat Hashcash. This is a completely bogus reason not to use Hashcash on your blog now. It is possible that one day hashcash will not be enoughto stop spambots. But at the moment the picture is far better for those using hashcash than it is for those relying on CAPTCHAs and Bayesian filtering. Make hay while the sun shines I say!
Samsung YP-Z5 portable MP3 music player review
17 September 2006 - 13:12
I've been wanting to buy a portable music player for years but been put off by a number of things. First was that I really wanted a flash-memory based player (rather than harddisk based) so that I could carry it while running, but I wasn't happy with the capacity offered by most models. Then Apple released the first iPod Nano with 4Gb of memory and as Apple have done so many times before, they changed the market. All of a sudden 1Gb was no longer the top capacity available in a flash-memory based player. However it's taken a long time for other manufacturers to catch up with Apple's iPod Nano.
Before I talk about the player I eventually bought, the Samsung YP-Z5, let me explain why I didn't just buy an iPod Nano. If I just wanted to rip my own CDs and maybe play a few mp3 files I *ahem* collected from other places, the iPod would have been a fine choice. But ever since all-you-can-download subscription services have been available (like Napster, Rhapsody, etc.) I've been quite taken with the idea of being able to fill up the player with whatever music I want, all for a flat fee. Sadly, Apple has not seen the light on subscription services and their iTunes Store only supports pay-per-track downloads. Due to all the incompatabilities that DRM (Digital Rights Management) introduces, that means all Apple players are out of consideration. The other big player in DRM is Microsoft and the good thing about Microsoft's Windows Media DRM is that it is licenced to many manufacturers so you get a wide choice of players rather than being locked into those of one manufacturer (Apple). It has to be said that there are very few products as good as Apple's, but I believe the YP-Z5 is one of them.[Read More]
Robert Scoble leaves Microsoft
12 June 2006 - 09:20
Microsoft's top blogger (or least their top blogging proponent) Robert Scoble has decided to leave the mega-corp and join a start-up specialising in podcasting and video blogging. Although his job at Microsoft was all-round image-enhancer, his main job as far as I was concerned was co-founding and producing a large proportion of the videos on the Channel 9 site. There's a lot of stuff flying around about Robert's move - he sets a lot of things straight with two long posts on his own blog and another post on Channel 9 explaining how this doesn't mean Channel 9 is over.
Personally, I think if this was anyone else leaving Microsoft, or Scoble leaving another other large company it wouldn't be nearly as big a deal. The anti-Microsoft mob are always going to look for some angle on how Microsoft's evil made it impossible for him to stay. The anti-blogging mob are going to look for some angle about how this means Microsoft doesn't really like staff bloggers after all. When it comes down to it, the company he's moving to is the perfect setup for Robert and if you had the chance to take the job that's perfect for you, you'd do it even if it meant taking a bit of flak in the meantime.
Scoble has been described as Microsoft's "Cheif humanising officer" and for sure, I feel better about Microsoft being a large group of people instead of the huge faceless corporation it seemed before. And I wasn't even in the anti-Microsoft mob to start with. Things have changed at Microsoft over the past few years and it's not all down to Scoble, but he does symbolise the changes that have occured in many departments. It used to be that if you found a bug in a product you'd have to go through support (if you had a support contract with them) or try emailing a general bug report address. You might get a response, you might not. You probably wouldn't get any communication from the people that actually developed the product that's for sure. Now you have developers blogging about what they're doing. They take feedback, some teams even give the great unwashed access to their bug tracking systems, and of course there are those crazy enough to appear on Channel 9 videos.
So good luck Robert. I hope it works out with the startup and it makes you rich! Microsoft will manage without you I feel and at least Channel 9 videos will no longer be punctuated by your manic laugh :)
Napster blames Micrsoft for it's own failings
07 March 2006 - 05:11
Napster got itself some publicity last week when their Chairman CEO Chris Gorog publically blamed Microsoft for their inability to dent the market share of Apple's iTunes Store. I posted previously about the Napster user experience. I don't doubt that a lot of Napster's failure is down to the failure of Creative and iRiver to make mp3 players that are as innovative and desireable as the various iPod models, but there's plenty Napster could do to improve their service. They don't have the right to sit there and say "we have this service that's so much better than iTunes, why is nobody using it?" First of all they need to understand that because they provide a subscription service they're not going to sell nearly as many tracks per-user as iTunes Store is. Secondly, they need to sort out the audio quality issue - 128kpbs WMA is not good enough when you are paying £0.79 for around 3 minutes of audio. Lastly there are a number of things in the client application that really need a revamp:-
- The interface uses multiple instances of the Internet Explorer control. The problem is that it switches between them almost instantly, then it loads the new content in. So you can hit the back or forwards buttons on the toolbar and get the incorrect pages appear before the correct pages are loaded. I don't know why they thought using multiple IE controls would be a good idea - maybe it works well on their LAN were the new pages get loaded intantaneously, but when you are on a slower internet connection, it sucks.
- What also sucks is that the main front page displays a "Working..." animation while it loads. I'm sure they thought it would look nice to have all the content appear at once in a flash, but it really slows things down massively. Browser writers put a lot of effort into making web pages render progressively as the content loads because it makes the whole experience more fluid. The Napster designers have seen fit to undo that work because they don't want to it become too apparent - at least on the first page - that they are using browser controls in their application.
- Quite often the web page part of the display that shows the album name, cover and related artists loads fine, but the grid-view below it where the tracks are supposed to be listed, doesn't get populated. I guess the data is comming from different servers and one of them isn't very reliable. Anyway, it sucks also.
- The recommendations system is fairly useless. They need to take a long look at Amazon.com's system. That managed to predict much of my CD collection after I'd listed about 10 albums. I've listened to countless albums on Napster now and it still recommends stuff that I've never listened to and never will listen to. Oh, and unlike Amazon, it doesn't have a "never show me this again" button for recommendations that are way off.
- The radio station system is broken as I detailed previously.
- The Napster catalog is too simplistic. It groups tracks into albums and assigns albums to artists. To be fair it does have the smarts to assign tracks to artists as well, so that when you search by artist you get the tracks they contributed to compilations as well. However there is no way to browse albums that are not by a particular artist. Searching for "various artists" doesn't get you far either.
- Another problem with the catalog is that every collection of tracks is an album. That includes CD singles and EPs. So you can get a band that's only a few years old with 20 to 30 albums and most of them are just multiple single releases. And it's not clear in the user interface which albums are real albums and which are just singles. They need to colour them differently or something and also add a checkbox to the page to filter out albums that aren't real albums.
There are probably lots of other things that are wrong with Napster as it stands, but another problem is the lack of marketing. Very few people I speak to know that Napster runs a legal download service now. It's also difficult explaining the subscription service. Napster really needs to do a better job of getting the word out if it wants customers.
Oh, and I'm posting all this to my blog because Napster is one of those companies that puts as much distance between itself and it's customers as possible. They have a "user feedback" form, but there's no conversation with that - they might read your comments, they might not, either way you're not getting a response from them. They have a forums system build into the application, but there are no staff to be seen there - it's just all 15 yearolds slagging each other off. Maybe I should recommend them a copy of Scoble's Naked Conversations.
The poor state of Formula 1 team websites
14 February 2006 - 17:27
Robert Scoble is famous for saying that if your marketing website doesn't have an RSS feed you should be fired. He's taken a lot of criticism for that stance and when I first read it I wasn't convinced it was that big a deal. However, I was recently compiling a list of Formula 1 team websites and was disappointed to find that only one of them had an RSS feed for it's news page. So I guess I'm now a believer.
For those not familiar with Formula 1 racing, there are currently 11 teams. All teams are required to design and manufacture their own cars (the engine may be bought in) to very high standards of both performance and safety. There are 18 races this year in North and South America, Europe, Asia and Oceania. Annual budgets range from $40M to $300M and staff levels from around 100 to over 1000. 6 of the 11 teams are owned by or closely allied to major motor car manufacturers. So this is a big global business - a long way from the likes of Champcar or IRL - and only the best will do for even the smallest of F1 operations. Even the manufacturer owned teams depend on external sponsorship and sponsors require maximum exposure in order to justify the money they hand over. That makes every chance to expose and promote their sponsors very important to the teams. So you'd think they would put a lot of effort into their web presence....
Well in a way they do. Almost all the sites are slickly produced by experienced web design agencies. Unfortunately these design agencies seem most intent on fulfilling a brief that was penned by someone who knows nothing about web usability or even good web marketing - as long as it looks pretty on this person's screen, they are happy.
The following is a table of woe. As you can see, only Renault have an RSS feed for the news section of their website. Many of the websites are all-flash affairs. Now I don't object to the use of flash on a website, but if the entire site is one big flash file it makes it impossible to bookmark pages, navigate using the standard back and forward buttons, or copy/paste from press releases etc. And that's without going into many of the accessability issues that web-standards fanatics like to go on about. Those sites that avoid the all-flash fate then have a chance to fail by using Frames is such a way as to make the bookmarking of individual pages almost impossible.
|Team Name (link)||RSS Feed||All Flash||Bookmarkable pages|
|Midland F1||-||-||X |
|Scuderia Toro Rosso||-||X||-|
|Toyota F1||-||-||X |
|Super Aguri F1||-||?||?|
Renault F1 is the only team to come out with full marks - fittingly for the 2005 world champions! Williams, Midland F1, Toyota F1 and Honda are almost as good, they just need to add an RSS feed so fans can keep up with their news more easily. The rest should hang their heads in shame, sack their current web partners and start again.
[Note: Super Aguri F1 haven't really got a website yet - they're got their work cut out getting a car ready for the start of the season being the new boys for 2006 - so I'm not rating them yet.]
Prepare for Cell Processor hype onslaught
08 February 2006 - 08:44
Just watching the US CNBC Squawkbox over here in England and they're going on about how amazingly powerful and revolutionary the new IBM Cell Processor is going to be. And if a product that isn't available yet and for which there is no new news available to get appearances on CNBC means that the PR engine has swung into full-on hype mode. Expect floods of blog postings about the chip and stories in all the news websites.
It's all PR hype though. We know the Cell is going to be the core of the Sony Playstaion 3 and there are possibly other applications for it in workstation accelerators. But don't expect to see it as widely used as IBM would like it to be. Firstly the Cell is not a cheap chip to manufacture. Some of the early hype made claims that by putting lots of simple processors together on one die they could make a very powerful chip at low cost compared to traditional solutions. This is just not the case though. The primary determiner in the manufacturing cost of chip is the die space it takes up and the Cell is as big as any high performance design. Even though the individual units of the Cell (SPE or Synergistic Processing Elements in IBM speak) are small and simple, when you jam 8 of them together (and add another true PowerPC processor to co-ordinate them) you end up with a big chip.
The other concern over the Cell is the ease of programming. Writing highly multi-threaded code is difficult anyway unless you have an easily seperable algorithm as your main computation task. But if you then take into account that the SPEs don't have access to main memory - only the 512KB they are bundled with - the whole business of co-ordinating data transfers to and from the SPEs adds an extra layer of complexity that any programmer would rather do without.
There is plenty of debate among game programmers over whether the PS3 offers as much useable processing power as the more conventional three-core PowerPC CPU in the Xbox 360 because the potention of the Cell will be so hard to realise. Being able to decode 48 MPEG2 streams simultaneously isn't that useful in most applications.
Wakeup call for Technorati/Feedburner
27 January 2006 - 12:03
I only recently got into the whole blogging lark. Although I've been knowlegable in the ways of HTML almost as long as I've been using the web (it all came along when I had mountains of spare time at university), I had to learn a whole load of new technology and 'customs' when I relaised that the best way to bring my website alive was to use blogging.[Read More]
Review of Napster Subscription Service
21 January 2006 - 14:54
My PC recently suffered yet another harddrive crash (I'm never buying another Maxtor drive) resulting in the loss of about half of my mp3 collection. Although I could rip my own CDs again, I have to admit there was a large number of files there that I, erm, don't have on CD. Since I was also temporarily down on HD space I thought I'd give a music subscription service a go. Napster is only £10 a month (and yet somehow it is also US$10 a month - how does that work?). They also have a free one-week trial.[Read More]
BMW gets Intel sponsorship (ironically)
15 December 2005 - 17:51The BMW F1 team today anounced that they have signed chip giant Intel as their title sponsor. There had been rumours that Intel was in discussions with several teams about a big-money sponsorship deal for some time. Bitter rival AMD have sponsored Ferrari for a few years now and it seemed about time such a large company as Intel would get involved. What makes this news ironic? Well... BMW is of course the new name for the Sauber Formula 1 team and with some fanfare they anounced in December 2004 that it had invested in a super-computer to run CFD (Computation Fluid Dynamics) simulations. The computer, named Albert (for Albert Einstein) is powered by, you guess it, AMD chips. 530 Opterons to be precise. There is no news as yet on whether this machine will be replaced by Intel hardware at the earliest opportunity.