May 2004 Archives

Lesson one: break your news feed.

Here is a good example of what not to do when changing the backend that produces your news feed, illustrated with OsOpinion. Here is how I was greeted today in my aggregator:

Mmm'kay, so since they don't provide the new feed URL in the text, let's visit the link as advised...

Ah! Very helpful. So I had to visit the site, to find that the feed was moved from http://www.osopinion.com/OSOlinks2.xml to http://www.osopinion.com/backend.php.

It's not the first time I see this, so here is just one advice you may consider to avoid losing those of your readers who will not care following your moving feeds: Cool URIs don't change! Changing the URL of a news feed is strictly equivalent to changing the URL of your home page, since the feed URL is actually the site entry point for its subscribers. Would you break your home page like this just because you've changed your content management system? No one in their sane mind would accept that a CMS breaks a home page like this, so why accept it for a news feed? OsOpinion could have changed their backend without changing the feed URL.

Would someone stop this guy from designing, designing and designing all over again?

And no, despite all their efforts to fool you, he's not the same as this one who's redesigning at the same exact moment with a rather provocative (therefore interesting) cut on comments.

What's up with this need for this web redesign fury, guys? Must be spring cleaning or something.

As if I needed a redesign.

Hum, what I need is, err, a design to start with ;-).

In all typical webmaster's geekiness, I regularly explore my web server logs to try to understand how people use this site, as well as detect anything bizarre such as referrer spam and other Internet weirdness.

For the referrer spam, I'm reluctant to post the URLs as this would favor their position in search engines, but here is the tutorial from Mark Pilgrim for Apache and which is the method I use block certain robots from doing their dirty work here. The way I manage it is very simple, I have defined a template in MT that points to a .htaccess file at the site's root, so I can update it easily with a browser.

Earlier in the year, I had banned AvantBrowser because it stubbornly pretended that their home page referred people to my site. This was an error in an old version of this browser which has been corrected as far as I can see. Very recently, I've started to become suspicious after the apparition of regular hits apparently referred from... "http://padawan.info". Because of an unresolved bug in the DNS service I use for this domain, there is no such thing as http://padawan.info because the DNS rejects all my attempts to define a record for this short version. So, I started to suspect a dirty robot, only to find out this in my logs (referrer highlighted):

<IP removed> - - [23/May/2004:14:19:19 +0200] "GET /culture/fahrenheit_911_gets_cannes_palme_dor.html HTTP/1.1" 200 20528 "http://padawan.info" "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2.1; aggregator:NewsMonster; http://www.newsmonster.org/) Gecko/20021130"

I explored further to reach the conclusion that not all NewsMonster installations out there do such stupid thing as to send a false referrer, since the only culprit seems to always come from the same IP belonging to the following range:

Syntrio SYNTRIO-66-180-224-0 (NET-66-180-224-0-1)

                                  66.180.224.0 - 66.180.255.255

So the innocent culprit can contact me if s/he recognize this range or the name "Metasearch Consulting Network".

Meanwhile I have sent an email to the developer to raise this simple issue: sending false referrer information is a sure way to be assimilated with dirty referrer spammers, and that surely is an embarrassment you do not want for your legitimate product, isn't it?

P.S. That wasn't long either. Here is Kevin Burton's reply:

That was fixed LONG ago... it's not my fault that people run old code ;)

So the innocent culprit at 66.180.236.something, please update your copy of NewsMonster!

But the morale of the story remains intact: it should not have been in the code in the first place!

I didn't plan to upgrade my Movable Type installation right now, but a spammer who's apparently in love with my site left about 160 comment-spams in a couple hours. The automatic IP ban throttle of MT 2.661 is a joke, I received a notification of ban after 80 comments! So I decided to test the new comments management features and voilà, upgrade plus comment deletion in less than 30 minutes.

One caveat though that I discovered during the process (I missed it during the alpha/beta test): the default encoding has been changed from iso-8859-1 to utf-8 and this is BAD for an upgrade. If your existing content contains non ASCII characters coded in iso-8859-1, you will need to change the PublishCharset preference as well as check that NoHTMLEntities has the right setting (for me it was commented out in 2.661, it's set to 1 in MT 3.0D). I'm glad I had a second user with a login name in ASCII, because I couldn't login after the upgrade with my regular login name because of the encoding! If your content is pure ASCII, you don't need to worry about this change.

I didn't change any template yet, If you experience anything funny with this weblog -- and I'm not refering to its content ;-) -- let me know.

Now is the time to think about which policy I'm going to adopt to keep this dumbass from spamming this site without annoying all legitimate commenters. I don't want to use TypeKey for various reasons (the first one being that I don't think it will deter spamming, the second one being that as a European citizen, I resent forcing visitors to leave their personal information on a central server hosted in a country that doesn't respect the European laws on privacy protection.) I'm also resisting closing comments after a certain time, since I've received interesting comments in the past on old entries, however it's probably the most efficient method and I might have to resort to it. A good method in MT 3 would be to enable moderation, but it requires a great reactivity on my side and, I fear, is rather aggressive in a world where people are used to instantaneity.

BTW, I set the comment throttle to 60 seconds, so if you plan to post two comments in a row, take some time to polish that second comment ;-).

News that part of the roof of terminal 2E at Paris Charles de Gaulle has collapsed over a boarding area of about 30m, at 7h this morning (the BBC reports that at least six people were killed and three hurt, which the radio currently confirms as probably definitive.) Flights (Air France / Sky Team alliance) from New York and Johannesburg had just arrived at the terminal. Passengers were also preparing to board a flight to Prague.

This terminal is the most recent one built at CDG, it was inaugurated on June 10, 2003 and its opening temporarily delayed (I recall it was because a light fell out from the vault during a security inspection) and finally opened on June 25. Its architect is Paul Andreu, currently building the great opera of Beijing (see "Grand Théâtre National de Chine" on this page.)

P.S.: apparently three of the dead victims were policemen who tried to establish a safety perimeter just before the collapse. P.P.S.: wrong report from the local media, those policemen escaped safe from the accident. There are four dead and three injured victims.

Or "Anti-Bush film tops Cannes awards" as the BBC puts it, or "'Fahrenheit 9/11': Connecting With a Hard Left" as the Washington Post puts it.

Funnily enough, Michael Moore's own web site is still sporting old news from before the ceremony. The man surely is overwhelmed by now ;-).

I can't resist to cite Bill Humpries and his What's a Pam Door?:

• Moore got more publicity.
• Quentin Tarantino, chair of this year's jury, got to throw a huge bomb.

• The chattering classes on the Right have been fed raw meat, and can now rant on about French perfidity. (Even if the international jury was chaired by an American.)

  I can't wait to read the Shorter DenBeste on this.

• The chattering classes on the Left count coup (see first paragraph.)

I can't wait to see this movie and, hopefully, Americans won't have to wait until after the election to see it too!

After explaining how I use MT personally, I have more thoughts to add to the conversation that I hope can find their way to San Mateo.

I approach MT with two very different angles: the first one is personal since I've started to use it in November 2001 and launched this very site with it 18 months ago. The second one is the corporate angle, precisely investigating the implementation of a weblog farm at Capgemini (I've listed my requirements in this post.)

When I saw the draft price list just before the public announcement, one with four digits figures on it for the high-end business licenses, I must confess that I did not foresee the public (track)backslash that quickly followed. But I could have, since I expressed to the Six Apart folks my lack of comfort with the (number of weblogs x number of authors) equation. The thing is, this formula is no more appealing to me as an individual than as a corporate user, but for different reasons.

As an individual power user, I've always resented software that have artificial barriers, artificial in the sense that they're applied on top of a unique codebase that, inherently, has no such limits. As a corporate buyer, my policy is very simply to pay a fair price for a concrete deliverable, i.e. I agree with suppliers on a certain amount of money vs. a certain amount of measurable work whatever its form (be it a good, a software or a service), and I think I'm a fair buyer to deal with. For example, I can definitely understand that support cannot come for free and that properly supporting all their users, from individuals to companies, is a core issue by itself for Six Apart, and a pretty complex one indeed. As a counter example, I told them that I would flat-out refuse the single-CPU limit since how I optimize my hardware infrastructure is entirely my business, not theirs (I turned down a well know database vendor for the same reason, but in its case this limit is not a copy/paste error in the license.)

I do understand why Six Apart made the decision to base its model on both the number of weblogs (as web sites) and authors. They're the two simplest variables they can easily measure, and they tried to avoid cluttering their price list by splitting this only according to two profiles: commercial and non commercial. However, this apparent simplicity might be one source of problem, because such model with fixed granularity cannot fit all their audiences and that the number of weblogs and authors are not orthogonal dimensions by which you can separate the modest, casual weblogger from the big corporation.

Here are the various target audiences I can think of, their differences and what kind of license could fly for them (your mileage may seriously vary!):

The individual weblogger

Profile: from casual diarist to serial blogger, usually of the power user type, at least one who prefers a custom solution that can be extended (hacks, plug-ins, etc.) to ASP services such as TypePad or Blogger.

# of weblogs/authors: highly variable and obviously a sensitive issue! Can be prolific, or host family and friends as a courtesy on their own server.

License: a personal license, free under a certain limit (with non priority access to the users forum), flat fee for unlimited weblogs/authors (and priority access to the support forum). $70 seems a fair price considering that MT 3.0 doesn't exactly lead the way in terms of innovative features (MT 3.1 might be an entirely different story.)

Rationale: word of mouth! The success of Six Apart is not only due to the quality of MT as an application, but also to the vast community of individuals who have embraced it. Word of mouth does not come for granted and is as volatile as the mouths owners. Why word of mouth is important for a weblog software: because if it isn't, what is in the blogosphere?

The web developer (freelance or employed)

Profile: the ones who will perceive MT as a platform and develop plug-ins or innovative frameworks they will sell to their clients to power their web sites including, but not limited to, weblogs.

# of weblogs/authors: absolutely senseless to them. Will have as many active weblogs and authors as active projects, and inactive as big as their portfolio.

License: a developer license at a price that accounts for the specific documentation and support that they require.

Rationale: they can discover the tool through the personal license and will pay if they understand the benefits of a developer program which I think is a required element to prove and ensure that MT really is a platform that's worth building on in the long run. MT developers will be the main sales people of Six Apart towards businesses.

The "non-profit" types

Profile: non-profit organizations, groups, communities, sharities, etc.

# of weblogs/authors: very tricky. Some community sites can live with a single website and an awful lot of authors with a high turnover. Some others will have many weblogs driven by a single author. Some weblogs and related authoring activities will be heavily event-based. This is an example where a linear curve (weblogs = authors) isn't granular enough and will be perceived as difficult to those in charge when comes the time to review the license fees as their usage varies.

License: a non-profit license that decouples weblogs/authors, sort out the complexity of finding out how many authors should be counted (e.g. flatten the number as the average active authors over a period of time) and provides a capped fee for which there are no limits.

Rationale: non-profits are interesting for word of mouth too and can provide an interesting test bed for innovative uses of weblogs. But they don't always have more money than individual webloggers.

The non-profit/public educational type

Profile: teacher, school, university, excluding professional training businesses.

# of weblogs/authors: senseless to them. Will have as many active weblogs and authors as courses and classes.

License: the educational license should be pretty close to the non-profit license, except they might even have less money and legal specificities I'm not aware of. May be both licenses could be merged into one educational/non-profit license.

Rationale: same as above, plus to expose potential users to weblogging!

The commercial license

Profile: all commercial users except the developers.

# of weblogs/authors: the current system may work for them, although it would benefit from being more flexible with a clarification on the installation, support and maintenance fees, plus a level at which big installations are licensed at a capped price + maintenance fee.

License: the commercial license which should be quite similar to the new 3.0 commercial license, accounting for the ones listed above.

Rationale: businesses will be the most sensitive to support and the chances of survival of suppliers over the long term (and by suppliers I mean both Six Apart and the developers here!) but they might be uncomfortable with a variable fee after a certain price point, especially at the beginning if they don't know where they're going. Starting with the weblogs/authors plans can help plant the weblogs in house, knowing that if they hit success, the costs can be capped.

So here I am, hanged on my personal and business naïveté, concluding that a more complex model is better :-). I clearly understand why Six Apart worries about some people installing MT to provide semi-industrial weblog hosting that will highjack TypePad or other offerings. MT is the only product to date that can do this pretty much out of the box and even makes it tempting. Being paranoid is not necessarily a drawback in business (look at Microsoft), but it's not a good idea to punish the whole community because of a few bad players. By breaking a monolithic model and rethink it according to each target audience, I think Six Apart may satisfy more of its constituencies that have not much in common apart their weblog tool.

Update 2: see John Gruber's page An Ounce of Prevention which is kept updated with the latest information.

Update 1: Apple has released a security update that fixes the flaw within the Help Viewer (released on May 21 but the fix is oddly dated 2004-05-24):

Security Update 2004-05-24 delivers a number of security enhancements and is recommended for all Macintosh users. This update includes the following components:

HelpViewer

--

Security firm Secunia has published a security advisory about a critical security flaw in Safari and IE 5.2 the Help Viewer which allows for the execution of scripts in the system with a simple URL, such as this one :

<a href="help:runscript=MacHelp.help/Contents/Resources/English.lproj/shrd/OpnApp.scpt string=usr:bin:top">click to run 'top'</a> (test for yourself if you're on Mac OS X, the following link will launch a Terminal window and execute the utility 'top' that shows the running processes: click to run 'top', just press 'q' to quit top then quit the Terminal and the Help Viewer.)

[Via MacMinute and CNet which says that Apple is aware of the issue. Code above from Simon Willison]

P.S.: if you're of the paranoid type you have a few solutions until Apple fixes this flaw:

• Install Don't Go There GURLFriend! from isophonic.net
• Install MoreInternet and map the "help:" URI handler to some harmless application such as Chess

"How are you using the tool?" asks Mena Trott to Movable Type users.

My own setup is as follows.

Padawan.info is built with three weblogs, one in English, one in French and one multilingual links weblog that I built to supplement the lack of such functionality in MT (which exists in TypePad under the name TypeLists). I consider these weblogs a logical assembly that makes a single web site, however I'm not sure that the recent changes in the license terms cover this setup, i.e. can this site be viewed as a single web site (as I view it) or are www.padawan.info and www.padawan.info/fr two different sites according to Six Apart?

I also have a test weblog that I use to test new templates, or hacks, or any modification I fancy but don't want to test live on the visible site. It's purely a sandbox for development purposes, and I would strongly resent if I had to count it as part of the number of active weblogs. But I think that this is not an issue with the present terms (although a clarification would help, especially for the developers.)

I have two author accounts just for myself to overcome a bug or a limitation in the MT API. My first account contains my first name (François) which has a diacritical character in it (ç). This worked well until I started to use desktop applications such as NetNewsWire or Ecto to post to my weblog. It seems that the API doesn't like to swallow a login name that contains diacriticals, so I had to create a new login name in pure ASCII to overcome this limitation (it also helps when I post from abroad on US/UK keyboards, since I can't get accents from them.) But I left the first account so that old posts made under that account are not impacted. I'm not sure that the license covers the case of the same person using different accounts (I can think of another reason, an editorial choice, where a single author would want to post under different avatars.) May be the license should be based on the number of identified individuals editing content, not as the number of authors accounts in an MT installation.

I also have a second domain (currently hidden) under the same server, but because of the way my hosting provider has configured it, I have two different accounts and am forced to install a second copy of MT. This has nothing to do with MT's capacities to handle multiple weblogs, just a server configuration issue that forces me to have two installations (therefore two licenses) when just one would suffice.

The fact is, after 18 months of using and sometimes hacking MT, that the way it works and the features it has (or has not) make me uncomfortable with both limits in terms of weblogs and authors, because I found that I had to leverage them to work around limitations or push the software forward.

More thoughts will follow soon.

D. Keith Robinson over at Asterisk has a great story on how non standard code hurts the bottom line and I think he's right to move the cursor from why web standards are good for business to why using crap code is more and more a terrible idea.

I particularly liked this bit:

It is an absolute mess. There are improperly nested tables, deprecated tags (font tags, small tags) inline styles, the works. You name it and it’s probably in there. This makes it very, very hard to customize, which is crazy because that’s what it was built for!

CSS anyone?

I’d spend hours just trying to sort through it all and then, when I feel like I’ve got enough of a handle to make a change, invariably I break something. It’s almost like they people who coded it wanted it to break.

I liked it because I've been through this myself, big time. To the point that I made the decision to scrap an entire section of my biggest corporate site during the recent rebranding of my company, because not only the code was horrible but to add insult to injury it was embedded into a complex mix of XSLT and JSP templates. Looking at ugly HTML code made of nested tables is not fun but when it's even more obfuscated into two extra layers of utterly ugly looking languages, it's more than I can take. I did not have the time to fix this mess and knowing that 95% of this stupid code was useless and an impediment to change, despite the early and repeated warnings I gave to its developers, made me quite angry but glad that I scrapped them for good.

We're still clearly in a transition phase, where more and more companies will join the bandwagon, but it's interesting to see the raise of the negative effects of sticking with bad habits vs. the benefits of adopting new rules as a driver for the adoption of web standards. Most companies never change direction until they hit the wall, as do many politicians who lack vision. It seems like a common trait of human nature.

I was sure that by not crying with the wolves and expressing a dissenting opinion against the vast majority of the MT-based blogosphere would attract some nasty comment, so here I am painted in a corner of my corporate ivory tower, draped in my corporate bias.

Well, I continue to think that the sky is not falling, that Six Apart is not an evil company, that developers deserve to eat and are not some sort of satanic cult.

In trying to list what I think are the important mistakes done by Six Apart, I forgot one: never under estimate the resistance to change. I'm being told that I forgot the history of MT and its community approach. Give me a break, please! I don't feel part of a community just because I put a "Powered by MT" link on my weblog or said nice things about how great MT is or how cool the Trotts are. I feel part of a community because I've done a certain amount of analysis about personal and business uses of such a weblog software, because I actively participated to alpha and beta tests, because I gave feedback to Six Apart, because I hacked MT and documented my hacks, because MT allowed me to build something that still amazes me today and, last but not least, because I paid for this software! It's been crystal clear to me from the start that Six Apart was an ambitious commercial endeavor, I've never assumed that I could make a free lunch from the entire Six Apart portfolio just because it's nice and cool.

I do understand why most people are angry though and their reactions have forced Six Apart to clarify and modify its approach. What I find surprising (and this is the reason of my own reaction) is how extreme their reactions are and how demanding, impatient -- and, to some degree, unfair -- they can be. Let me give you one good example of that. In all those posts about how "Six Apart sucks and I'm going to switch to X/Y/Z tomorrow morning", how many are reminding that Six Apart has always committed to providing means to export all your data in order to help you switch to another weblog software should you feel the need to?

I'm confident that all of this will actually lead to even more momentum to the blogosphere. It raises the profile of many very good free and open source software, competition is well alive and kicking -- pMachine is giving away 1000 free licenses of its commercial product ExpressionEngine (but hurry up, offers ends tonight, and be aware that it's a one-time deal only, upgrades aren't likely to be free) -- but most importantly, as Brad Choate notes, Movable Type is moving again:

I’ve been sad for a while about the lack of movement and development of Movable Type. True, the 3.0 release is less about features and more about empowerment— there are far more hooks into Movable Type for us developers to make use of. TypePad has been getting a lot of nice features for some time now. Is it coincidental that TypePad brings money into the company where Movable Type (by and large) has not?>

With Movable Type 3 providing a revenue stream for the company, it will bring resources to Six Apart to continue development of the product. That’s what users want, more than anything.

I for one am looking forward to the innovations that will come with Movable Type 3.

Me too, it's definitely not the time to miss the show.

One thing I've learned over the years, especially in the corporate world and even more in the blogosphere, is that when tempted to rant about something or flaming someone, one should wait and digest the facts rather than go up in arms instantly. Possibly the best advice I got from a colleague about that, when I can't wait, is to start writing right away the "How dare you, you little piece of..." then take a deep breath and rewrite in a "Dear colleague, thanks for your valuable input..." style.

It seems that Six Apart's announcement of Movable Type 3.0 Developer Edition exemplifies admirably this overreaction problem that many have, and that goes out of control thanks to the echo chamber amplification from TrackBacks on steroids.

Two days after this announcement, some sensible, constructive comments are starting to emerge from the wolves cries, from people who have taken the time to digest the news and understand what's really going on here:

1. MT3.0D is a developer edition. Most of the changes done in this version are related to the core engine and its API and will be invisible to the end-user until developers start to leverage it. Jay Allen, the developer of MT-Blacklist, nails it very well. Six Apart is turning MT into a development platform and for that, it needs developers. And for developers to start developing plugins, they need a development platform. Chicken and egg... If you're not a developer and MT2.661 satisfies your current needs, why rushing for a version that's not for you? Wait until this platform starts showing some new benefits, they will come, along with a feature release, probably MT 3.1.
2. Six Apart is a growing business. It's really amazing to see, still now, so many people who think that MT is a freeware -- some even believe it's an open source software! -- written by two cute bunnies called Ben and Mena who will work day and night to satisfy their wildest desiderata. Well, the reality is that the 40 or so employees of Six Apart cannot support everybody for free while fighting to establish themselves as a leader in their field, and a pretty competitive one indeed. It's also a start-up that needs time to manage its growth and start stabilizing its business.
3. Six Apart addresses very different users with different products. Almost a year ago, I did a little analysis on how Six Apart is building a weblog empire, which I think remains valid (including the MT Pro bit although under a different form). Individual users are served through MT 2.661, the free MT 3.0 and TypePad. Power users and developers are served with MT 3.0. Companies are served with TypePad, MT 3.0 and special licensing deals negotiated on a case by case basis. It seems to me that most individual bloggers have forgotten that, notably those who describe the new license as more restrictive, while business users (including developers) will find it less restrictive than the previous one.

Does this mean Six Apart did everything right? No. The padawan realizes that he still spends infinitely more time learning from his and others mistakes than giving lessons to anyone, but here are the few things I humbly think Six Apart could have done better:

• Never raise your customers expectations if you are not going to deliver soon (aka: under-promise, over-deliver.) This lesson I've learned at Apple when Jobs came back and stopped the bad habit of the company to pre-announce products a long, long time in advance, disappointing everyone and their dog who then had plenty of time to fantasize over a dream in their head that would never match what was eventually delivered (if delivered at all). Six Apart should not have raised our expectations with two things: the promise that MT would remain both free and on pair with TypePad, and Movable Type Pro (which I feel entitled to call a vaporware for now).
• Don't copy/paste broken business models. The license published on May 13 did contain a restriction that MT should be used solely on a mono-processor server (note: the revision done on the 15th has removed this restriction.) As a business buyer, I've always been reluctant to such restrictions, which are more typical of high-end databases than weblog software. I do not accept to pay for variables that are purely linked to my own performance (e.g. optimization of server resources) but I'm definitely willing to pay for concrete work done by a supplier (a license and maintenance fee for software development and upgrades, a support fee for support, etc.) -- the only variable that I can accept to be based purely on my sole performance is called a bonus pay!
• Don't mix all your audiences together. This one is tricky, precisely because of the weblog-centric nature of Six Apart. I don't think that they are bad at communicating with their users, as some have written here or there, but that their weblog makes it difficult to address different audiences specifically. IMHO, all their sites could benefit from a little information re-architecture to paint the picture distinctly to the individuals, the developers, the business users -- this might be a little controversial, but I'm on both sides and I can see that "L'Internationale de la blogosphère" is not ready to fly yet, i.e. you don't sell the same concepts and dreams in the same terms to "free lunch Joe Blogger" and to the BBC's CIO. Or may be I'm completely wrong, may be the CIOs are more mature and that's those individual bloggers who aren't getting it ;-).
• Granularity is important. The price list has figures with two to four digits (including the one that some corporate users got to see). Six Apart chose to index the price on two variables: the number of authors and the number of weblogs. I think the present model is not granular enough, too steep for individuals (I'm already using four weblogs and two authors to run this web site because of MT's present limitations!) and lacking clarity and flexibility for the high end corporate user. It could benefit from being clearly split in three: a limited but free version for individual/personal use, a flat-fee unlimited version under $70 for the personal use of power-users (like Jason Kottke suggests) and a granular price list for business users that should do a better job at explaining the options, notably in terms of support.

Now what?

Six Apart is listening and has already responded to many legitimate questions: expanding the free version limits, adding cheaper personal options, clarifying the notion of web site vs. weblog, removing the server processor limit, confirming the terms of the 2.661 license.

I think now that it is time to watch those for whom this version of MT is meant: the developers. For they are those who will make the difference between MT and other weblog software. As a technologist it is my role to evaluate different tools on an ongoing basis, in order to know a market and pick the best options for my company. And I would find it foolish to ditch MT altogether at a time where it's clearly opening new possibilities!

Now if only I could retrieve that second MT Key I got for my $45...

I'm terrible at reporting about events, so let's do it in images instead.

It starts with a badge:

My Internet life in London has always been a frustrating experience, and as counter-intuitive as it might seem, being the top web gun of one of the "big 5" IT companies doesn't really help.

I thought I could find a hotel equipped with WiFi but I found none that was both matching our expenses limits and be located where I wanted (Leicester Square). Lodging in London is very expensive and yet, it seems that few hotels have thought about the value that such a service could bring them.

I finally resorted to go to... Starbucks! I need to tell you that I avoid Starbucks like plague. Or like McDonalds, which is the same. But, hell, Starbucks was the only place in sight where I could hope to get a decent Internet connection through their hotspots. So I changed my French eating habits and took the plunge to the anglo-saxon utilitarian approach of lunch -- aka junk food quickly swallowed in front of a notebook -- at the Starbucks at Leicester Square.

The WiFi service of Starbucks in London is provided by the German operator T-Mobile, which sells the connection at the awfully expensive rate of £5 an hour. T-Mobile customers can "benefit" from a very special price of £1.5/15mn, which means £6/h (the reward for being a loyal customer I guess). I bought one hour with my VISA card (they don't accept Amex) and received my credentials (two mind boggling long chains of random characters) on screen. I logged in and started to enjoy The Web at last. After 20 mn, my joy faded as the connection dropped. The WiFi signal was present and strong, but I couldn't access any site. After fiddling for a few minutes, I decided to hit the logout button in that little T-Mobile popup window, the request to the T-Mobile site seemed to work and the popup window closed itself which I think I'm right to interpret as a positive sign of a successful logout). I quietly headed towards Hyde Park for a meeting. I got back to Leicester Square and before getting my luggage at the hotel, got back to Starbucks, decided to use my hour until the last minute. No luck, I was greeted with an "invalid password" error message, which I suppose is the IT equivalent of telling me that my hour had expired. Very nice. It meant that the logout request didn't come through because the network was down!

I browsed the few pages that T-Mobile allows you to see for free, which contain their terms of service, a F.A.Q and some troubleshooting instructions. Written by a lawyer in lawyer's language for most of them, those pages make it very clear that whatever happens, it's not anybody's fault but yours. Even the troubleshooting pages are written to make you feel like a hopeless form of life and one of the most useful piece of advice is that you should call your IT administrator if you're trying to configure your network settings on Windows NT (from a Starbucks?). And nowhere in this academic example of bad customer service one can find a way to contact someone at T-Mobiel. No contact form, no email, no nothing. We told you, if it doesn't work, it's your fault, so why would you dare to contact us?

The first lesson of the story is that unless I have no other choice between that and death, it will be long before I'm a customer of T-Mobile again, and one of Starbucks as a corollary. The second lesson is that those chains have not realized that you can setup a hotspot for about 100€, hook it to the internet connection that you most probably already have to run your business but are not using during open hours and offer the access to your paying customers for free. Then, why would one go to Starbucks when there are nicer places around that don't try to make a quick buck at every occasion with bad customer service?

I'm sure such places already exist. If a charitable soul could just tell me where they are, the padawan will be very grateful!

Off to London two days for a series of meetings and... this! Can't wait to meet with you guys.

Dave Pollard has published an excellent weblog functionality scorecard. He's missing at least one thing that's core to the web: the hyperlinks pointing back to one's weblog, in other words one's blogosphere (what Technorati calls Cosmos), etc.

I note with pleasure that the only feature he would be able to explain to his father over the phone, is WYSIWYG text editing and publishing. Enough with WYSIFUC!

Playing more with ExpressionEngine and digging its User Support Forums helped me figure out some information that's worth sharing, since it impacts my comparison between Movable Type and ExpressionEngine.

The main point is that I don't think EE is ready to work for a weblog farm. While the site and documentation indicate that you can have multiple weblogs, the caveats are:

• creating a new weblog does not create any template for it. It's a manual process and, as far as I can see, a rather cumbersome one. The default templates are capable of handling only the default weblog or all weblogs, but not a particular weblog without making them specific to it(*)
• there is no way to selectively limit access to a set of templates, typically allowing weblogs owners to modify their own templates
• the administration interface will not scale as the list of templates grows, since they are all displayed on one page! (a classic of UI scalability)

I understand that pMachine is working on a "user weblogs" or "community weblogs" module that would allow this, but there is no indication on when this will available.

This said, the claim that EE can manage multiple weblogs holds true. Simply it must be clear that templates have to be created manually for each weblog and managed centrally by the administrators, without an easy way to delegate this task.

For the technically inclined, here is, IMHO, the main difference between MT and EE when it comes to handling multiple weblogs:

Movable Type

MT templates are always executed within the context of one weblog at the time of (re)building pages, so there is no ambiguity here and, except for the case below, they do not contain any weblog-specific identification. This brings a huge benefit to templates: they can be duplicated very easily from one weblog to another without modification

Some tags (by default or from plugins) can address other weblogs (by inclusion or exclusion), in which case they point to weblogs by either their ID or their name

ExpressionEngine

EE templates are invoked by the request URL dynamically at view time, but the system has no way of determining from the URL if a particular weblog is being requested

EE tags refer to weblogs by their name. You have to specifically hard-code the weblog(s) name(s) in every template when you want to handle a specific weblog, and I have found no way to pass that name dynamically to the template. When the weblog name information is missing, the template will use the default weblog (or all of them in certain cases)

So my conclusion is that if your goal is to quickly setup an installation that will drive tens or hundreds of weblogs, this is something MT has been designed to do from the start and I would not recommend EE until its users weblogs module is released.

(*) At least I did not find how to do it. If I'm wrong in assuming that, please correct me.

P.S. and now that MT3.0 Dev Ed is out with its licensing conditions, I guess I'll have to update my review soon :-).

I'm in the process of evaluating several weblog solutions for building a corporate weblogs farm. In other words, I'm looking for a solution that would allow a company, with a simple (if not single) software installment on its existing web platform, to host any number of internal and/or external weblogs. My requirements and constraints for such a product are:

• Hosting on a LAMP platform: Linux, Apache, MySQL and PHP/Perl
• Ability to integrate the system to the company own authentication system
• Product installation and configuration support
• Maintenance (software updates and upgrade path)
• Produces pages that conform to the XHTML/CSS web standards
• Authors can author content with their browser of choice on either a PC or a Mac (with a preference for "modern" versions of IE on Windows, Safari on Mac OS X and Mozilla on both)

With this goals and constraints in mind, I ended up comparing Six Apart's Movable Type 3 and pMachine's ExpressionEngine 1.0. The following is not a full fledged comparison of both products, but I thought it might be of interest for other people as I came out with some interesting differences that might help someone else make their mind about them.

Some reminders and disclaimers before proceeding:

• always take others' views with an appropriate amount of salt and with respect to your own requirements
• this comparison is subjective, at least to my requirements above
• I've been using Movable Type for one and a half year, ExpressionEngine for four days
• this review is not extensive, I concentrated on the differences that matter to me and would help me determine which product is better suited for which usage. If you're looking for a "X is good, Y sucks" review, you'll be disappointed — I don't have for habit to include products that suck in my shortlists

As usual, I'm looking forward to reading your feedback to this review.

I'm working on a comparison between two leading weblog software and since it's already quite long, I'll be venting something in advance to free some space up and not pollute the spirit of the upcoming article.

I would like to know how you, software editors, dare to sell products that you tag as "the world's most creative and powerful web publishing software" or "the most powerful and customizable weblog publishing system available today" (you'll recognize yourselves) and still ship them with ridiculous, sometimes non cross-platform, WYSIFUC interfaces.

Come on, it's 2004 and no one has been able to get a WYSIWYG rich text editor in a tool that is supposed to help non technical people publish on the web?

And please, don't tell me about Markdown or Textile or -- pitié ! -- Wiki syntax. If you want to shove FUC on me, let me at least choose one that's a standard with some chance to survive the next decade.

I'm not asking for the moon, the cure for cancer or even a solution against Microsoft's monopoly. I'm asking for a simple interface, with the typical buttons, that would help those bozos people from marketing author their texts properly, with links, headers, blockquotes, images, even tables and any semantic that is permitted under the auspices of XHTML without having to type, see or even think about f*cking ugly code!

When you'll understand this, and get rid of the redundant preview button, you'll really deserve your superlatives. And I'll be a super happy webmaster. Thanks for listening.

Previous rants on same subject.

P.S.: I have to pay tribute to Stéphane Le Solliec for the acronym WYSIFUC. Thanks Stéphane, the conversation on that day was certainly great fun ;-).

The next big life project for the better half and me: move to San Francisco. Seriously. I'm chasing for advices... and a job, eventually!

Apple's teasing its developers:

You won't want to miss the Apple Worldwide Developers Conference (WWDC 2004) in San Francisco, where CEO Steve Jobs will unveil "Tiger," the next release of Mac OS X -- Apple's critically acclaimed operating system. WWDC is your opportunity to preview this new release and discover how Tiger will fuel the next generation of developer innovation.

Critically acclaimed? Is it because English remains a foreign language to me or is this words association really weird? However I'm looking forward to reading what Tiger is all about (I'd love to come but I'm afraid I won't be able to this year again.) I bet Steve will joke about the competition and its next generation to come in, what, 2006?

Tim Bray publishes Sun policy on public disclosure, a set of advices for Sun webloggers, along with a note on making this policy. Under "Tools" I noted that Sun has started to work on weblog software:

We're starting to develop tools to make it easy for anyone to start publishing, but if you feel the urge, don't wait for us; there are lots of decent blogging tools and hosts out there.

Almost a year ago I decided to modify my news feeds to carry the full posts instead of summaries. In order to do that, I changed the following line in all my feeds templates:

<description><$MTEntryExcerpt remove_html="1" encode_xml="1"$></description>

into:

<description><$MTEntryBody encode_xml="1"$></description>

One of my reader wrote that it was The Right Thing™ to do, the other one did not complain. And I lived happily ever after.

By doing that, I didn't realize that I made a mistake. If I wrote an entry with an extended body, the extended body wouldn't show up in the feed. Actually, I thought it was good, so that was not a mistake after all. And I started using this trick when writing certain entries, almost articles, that I deemed too long to go in extenso on either the home page or the feeds.

What I'm realizing all of a sudden is that now that my feeds subscribers are used to consume the full posts in their aggregators (and why would they come here on this ugly site anyway?), they are more likely to miss those bigger posts -- of course I leave clues in the summary that there is more to read, but I think that reading news with an aggregator puts one in a mode of fast skimming rather than reading long articles and one might miss it.

However, I wanted to keep this option to decide when a full post is not a full post after all. All I had to do was to replicate the way it works on my home page, i.e. modify one line in each feeds template:

<description><$MTEntryBody encode_xml="1"$></description>

and change it to:

<description><$MTEntryBody encode_xml="1"$><MTEntryIfExtended><![CDATA[<p><a href="<$MTEntryLink encode_xml="1"$>">more...</a></p>]]></MTEntryIfExtended></description>

Et voilà, now all my feeds will sport a "more..." link to the full post when there is more to read than meet the eyes!

Since I've sold the idea of weblogs within my company, I've been charged with the task to explain what are weblogs to people who haven't heard of them yet. I wrote this primer as a starter. This is a work in progress, with probably more to come. I'd love to have your feedback on it.

Danny Gregory made a moving post and drawing of the building at 90 West Street bordering Ground Zero in New York City :

I was riding my bike down the West Side yesterday afternoon and passed Ground Zero. It's a big construction site these days and, like a typical New Yorker, I just breezed past.
For some reason, this time I noticed the West Street Building on the south west corner and I stopped. I looked at it and I saw it for the first time. It's a landmark building, built in 1905 by Cass Gilbert who also designed my all-time downtown favorite, the Woolworth Building.
While all of the modern buildings round the site are either gone, rebuilt or heavily shrouded, the West Street building was openly wounded. Its Parisian mansard roof is completely draped in black steel mesh. Large pieces of its limestone facade are smashed or cracked off. Its terra cotta tiles, installed for fireproofing, helped to protect it from the burning columns that fell off 2 WTC but took a beating. Ornamental busts around the front door were decapitated. Through the empty windows I could see rubble in what once an elegant interior.
This building was so stately and built to endure. Now, it stands with gaping holes. My instinctive reaction was an angry sadness that the people who did this knew nothing about our city, didn't understand the significance of the history they erased. Not that it would have influenced them.

I noticed this building too when I went there last March, for the first time after 9/11. Actually it's rather hard to miss for it is so different from all other buildings around. Or may be it was my parisian eyes and love for architecture, facing the painful contrast of the images pre-9/11 I had of the WTC area in my memory and the giant open wound that Ground Zero still is.

I made a lot of pictures back then but didn't feel like posting them. But after seeing Danny's drawing, and even so this is quite prosaic, I'd like to share them.

You know its scary when legions of geeks are overwhelmingly against a new form of technology...

Slashdot's wisdom about California Secretary of State decision to ban certain Electronic Voting Systems from Diebold in three CA counties and threat to extend the ban to ten more counties. California will impose a paper voting trail to all machines in the state by 2006.

At the same time in Europe, an independent commission in Ireland recommends to suspend the use of the Nedap Powervote for the upcoming European elections next month (report, via The Register) because of serious doubts about their accuracy and secrecy. Meanwhile, despite all lessons learned all around the world about the flaws of such systems, the French administration quietly continues the roll out of the same Nedap systems that were introduced in one city in France for the March elections and will be extended to tens more for the June elections.

One month ago, the Open Voting Consortium demoed a free voting software that runs on very inexpensive PC hardware with an open source software and a tamper-proof, blind-friendly, voter-verifiable paper trail. The Mercury News called it the touch-screen holy grail, "An electronic voting system that's cheap, secure, accurate and easy to use. One that uses off-the-shelf hardware and publicly examinable software. One that voters can trust."