Somehow, it all adds up to very little output on this blog. Each outbound channel is its own context and I differentiate them for different uses and styles. This blog is the most valuable to me, so I suffer from terrible inertia when scribbling to it. Recently I've only been grabbing the MT posting interface when a post enters my head fully-formed and bursting for release. As ever, I have about twenty really-good half-formed ones that have been sitting there for years. (Fortunately for me, less fortunately for you, this is one of the half-formed ones.)

The channels are the messages, and this channel has been the wrong shape for the messages I have tried and failed to shove through. The channel won't change, so I need to change the way I form the messages. One day I'll realise that in practice; this is another attempt.

Based on recent experience doing more extended web dev work than I have for a while, I propose the acronym TPTOTA (pronounced tip-toe-ta), which stands for They've Probably Thought Of That Already. It is a virtuous consequence of decent API design by those who are actually using their API in plenty of everyday practical work, and extend their API design to solve regularly-encountered problems without compromising the API's existing clean lines. It means that when you (the API user) encounter a fairly common task or problem in the API's subject domain, these two lovely things happen:

1. You figure that the API designers have Probably Thought Of This Already, and you're right
2. You go looking in the single most obvious place for the solution, and it's there

... both of which allow you to relax with an Ovaltiney sigh of relief, solve the given problem in one swift move, and spend the rest of the day playing Dicewars.

For a perfect example, take a look at the Dojo Toolkit for Javascript. I was using Dojo's very lovely io.bind for some Ajaxy to-ing and fro-ing with the server, upon which I hit a snag with the way IE rudely insists on caching the output of most of those calls. I wondered if there was some simple technique for bypassing this problem, and a single Google search was all it took to show that it really couldn't be much simpler. Dojo is still a way from perfect, and the documentation reflects the immaturity of the rest of the package, but so far it's really terribly nice to work with.

Beware, however, of APIs which loudly proclaim their TPTOTA-ness yet leave you hanging. A great example: PHP. When talking to friends of mine who are fans of the language, I'd often hear praise along the lines of: "If there's a basic bit of code you need, you just go looking, and the chances are it's built-in already!" Well, no. Maybe I've just been incredibly unlucky, but in such cases I usually end up spending hours looking through a ludicrously-overgrown pile of inconsistently-named-and-signatured functions to turn up sod all. Case in point: I wanted to remove all null/zero values from an array. (At least, PHP calls it an array. I call it a hash, in more than one sense.) There are built-in array functions numbered like unto a billion. Half of them have a name starting with array_. Half (a different half, but not entirely different) take an array as first argument and single value as second, with the other half taking the reverse. To predict which order a function will use, flip a coin; it's about as reliable a method as any.

Presumably this is all to help users write as little code as possible, but I'd rather they made it easier to write the stuff they don't already have functions for. In order to perform the described operation, I used (as you would in most other languages) the filter - sorry, the array_filter function. And how to you provide the custom value-testing code? No, not as a function pointer or an anonymous function or a pure code block: you do it by creating a new function separately and then supplying the function's name in a string. Hey, why don't we just cut the compiler's balls off while we're at it? (After some exploration it turns out that you can supply an anonymous function with a special command that - and I am not making this up - takes the entire supplied function code as a string. Oh, PHP just loves eval, it's the fast-and-loose hot playmate that it runs around with instead of that staid old compiler who just complained about such behaviour and kept wanting to know what the language was actually, like, doing. GAH.)

And breathe.

Just so you can be sure: I've had a postponed blog entry cooking for, oooh, a couple of years now about why PHP has thoroughly beaten Perl in the web development marketplace, and all the things it got right that Perl didn't. PHP is, for most, a perfectly usable language that gets stuff done and with which you can sling web apps together pretty fast, and most of the time I get on with it just fine. It's just that the committee-designed car-crash illustrated above, like some others, is the kind of stain that makes me throw things and scream.

After much frustrating baby-triggered cancellation, circumstances have at last permitted me to deliver a proper talk - even if it was only seven minutes long. Tom Carden & Steve Coast's Techa Kucha Ask Later gathering was lots of fun: a kind of open-mic night for tech talks, with people running onto stage with 400 seconds to present the card stacks they'd mailed to Steve the night before. Lots of really good bite-size presentations, my favourites coming from Toms Carden and Armitage, though the strongest reaction was to a talk about Sudoku-solving in Ruby that included a web-crawler and home-built OCR engine.

Since I'm pretty happy with it, I've recorded the talk and slides (now with added demos) as a seven-minute Flash screencast: "Get Your Own!" The Build-To-Clone Design Pattern. The talk discusses the concept of software cloning and how it opens up new kinds of web applications. (I had been hoping to cover this as part of my tragically-cancelled Reboot talk, which - taking a cue from the latest trends in the games industry - I'm now hoping to deliver episodically.) I discuss the talk and the Timeliner app I created for it in more depth in this entry on the Ning Blog.

The whole thing's particularly timely as, the day after I presented, Tim O'Reilly mentioned Ning's cloning features during his keynote at OSCON. Speaking of which, that's where Ning PHP Deity David Sklar delivered his deliciously-titled I'm 200, You're 200: Codependency in the Age of the Mashup (PDF). It provides some excellent answers to questions I've had about the use of web services since they first arrived, so I strongly recommend it. Also you may note that we both, with no pre-agreement, used what is rapidly becoming the Ning standard sign-off. I wonder where we got that idea...

DAVELEY: I have this little dream whereby there's this whole village of reanimated corpses, and if you like, a kind of control tower at the centre of that village with a bank of monitors, and I control all the corpses.

WINTERGREEN: Why use corpses? Why not normal people? Why don't you just leave things the way they are?

DAVELEY: Because... because normal people... because I wouldn't have my tower! I want a tower.

Steve Coogan and Rebecca Front, The Day Today

He'd wanted to create something that would evolve. He'd hoped for a surprising pattern, an outcome not programmed, an unexpected turn of events, like the lovely life-forms that had emerged from Conway's world. Each time he brought the simulation back onto the screen, he'd have a moment of jittery anticipation. Maybe this time he'd see a leap. Maybe this would be the day when he'd bring the program out of the machine's internals to find a self-directing universe, a world that ran itself without the hand of the programmer. But except for the bug that once wiped the screen clean, it was always as it was now: a dull, repetitive place, a universe created by a not very imaginative God.

Ellen Ullman, The Bug

A story is told of several Rabbis, arguing over an abstruse point of law. One of them, Rabbi Eliezer, vehemently disagreed with the other sages. After long debate, he at last said, "If the law is as I say, may this carob tree prove it!" And the carob tree uprooted itself from its place. But the sages said, "No proof can be brought from the carob tree."

And Rabbi Eliezer said, "If the law is as I say, may the walls of the study house prove it!" And the walls of the study house began to bend inwards. But Rabbi Joshua rebuked them, saying, "When the sages debate, what right have you to interfere?" So, out of respect for Rabbi Joshua, the walls did not fall, but out of respect for Rabbi Eliezer they did not return to their place; hence they are still bent to this day.

And Rabbi Eliezer said, "If the law is as I say, may Heaven prove it!" And a voice came from Heaven, saying, "Why do you disagree with Rabbi Eliezer, seeing the law is always as he says?" And Rabbi Joshua stood up and said, "It is not in Heaven! It is not for a divine voice to decide the law, for in the Torah it is written that the majority opinion shall prevail." And the sages followed the majority opinion in their ruling, and not the opinion of Rabbi Eliezer.

And from this we learn that we are not to look to Heaven to solve the difficulties of our lives; that we are not to interpret signs and wonders to live our lives by them. We learn that there is value in making our own choices, even if God Himself communicates clearly that the choices we make are wrong. We learn that we may argue with God, that we may disobey His direct commandments and yet delight Him with our actions. We learn of God's compassion for us; in the end, broader than we can understand.

We read that, later, Rabbi Nathan met the prophet Elijah in a dream. And he said to the prophet, "What did the Almighty do, when Rabbi Joshua said, 'It is not in Heaven!'?" And Elijah replied, "At that moment, God laughed with joy, saying, 'My children have defeated Me, My children have defeated Me.'"

Naomi Alderman, Disobedience, quoting the Talmud (Tractate Bava Metzia)

Those of us who create giant, complex new worlds - worlds both totally imaginary and partially real - are often seen as megalomaniacs, control freaks, people who "want their towers". There is a germ of truth in this, but not nearly as much as there is in the total opposite: we also want our creations to be out of control. We want to create something that grows far bigger and wilder than we could ever be, than we could ever imagine, that leaves us merely gasping in its wake. We don't want to specify down to the last detail and be permanently at the controls; we want to create the tiniest seed and then let go, just watch. We want pride, but more than that, we want astonishment.

My son was born yesterday at 9:24pm. I don't know what he will grow to be. I will try to guide him and give him everything I can, but I am under no illusions about my ability to fine-tune a volcano. I wish for him to have the wildest dreams, and have the desire and ability to chase them. I may have other desires and hopes for him along the way, but the most important of them all is that he be able to choose for himself. All I can do is help him in every way I can. Every time he astonishes and surprises me, as I'm sure he will, it will make me happier than anything else.

If you want to make God laugh, tell him your plans.

In better news, there's a good chance I may be presenting at Reboot in Copenhagen on the 1st of July. Not certain about this yet, but the topic will likely be "Logic To The People" and will tie together Ning, Second Life, JotSpot and LambdaMOO amongst others. That's what I hope, anyway.

So, once again, my apologies to everyone in at XTech, especially the magnificent Edd Dumbill who has put up with endless vacillation and dithering from me on all kinds of topics. Oh, and the reason I couldn't make it? Well, obviously this is what I was anticipating when I cancelled the talk yesterday, but... bizarrely, while I was halfway through typing this very blog entry... my wife's just gone into labour.

(I'd better get off the computer.)

...

"And that one too."

(Took two old machines for recycling today, one of which was the original home of Shooting People, and had four 9GB SCSI drives precariously balanced in it. Should I ever get around to actually making us a server, it'll probably be on a quad-core 12GHz Xeon with 8GB RAM that my mum doesn't want any more.)

So, after many years of trying and failing, I finally made it to ETech. Hell yes it was worth the wait (since you ask) and I got to present at two sessions - one of them our own (for which I must thank David for co-presenting so ably), the other a five-minute slot in the microformats talk (for which I must thank Tantek (again)). Anyway, if you get the chance please do check out our session - not only is it summarised neatly with useful links in that thar page, but there's a screencast of the whole talk, which should answer most of the questions that most people fling at me about Ning. (Especially the "Can you explain Ning properly and give me some examples of how I'd use it? But hurry, I've only got 48 minutes and 51 seconds" one.)

As for the rest of it...

• lots of lovely chilled peeps
• debauchery and post-debauchery
• game of the conference: Werewolf or Animal Crossing?
• who can know what machinations churn in the mind of Shirky?
• Maker Faire Lite: battling Roombas, Atari VCS casemods and Esther Dyson firing marshmallows at everyone, hence gags about treading Dyson Spheres into the carpet. (We'd have cleaned them up, but the only vacuum cleaners available were busy fighting)

blonk (v.)

To blog without notable creativity, inspiration or merit; covering the same ground trod by countless others in the echo chamber; blogging as an alternative to thinking.
"I was going to write a considered piece about climate change in sub-Saharan Africa, but I've just been blonking pictures of my cat."

(Not deliberately invoking the Mornington Crescent exclamation, but not totally unrelated either, if one considers the infinite space in which we play this game as a giant board, with 80% of the players continually shunting into each other on the Just Quoting an A-Lister square. Or, for that matter, on the Making Up New Words about Blogging square.)

1. "Basic tracking works, but acceleration, scrolling, and right-click support are not available in Windows." (my emphasis). No right-clickee, no laundry. Yes, you can plug in a mouse, but who regularly uses a mouse when they have a trackpad handy? Come back when you've got a two-button trackpad and then I'll think about a MacBook.
2. Remember the lovely head-in-sand days of Apple trying to sell OS9 on G3s by showing how much faster Photoshop filters were than on the equivalent PC? ("If only," Tim said while trying to port Starship Titanic to the Mac, "one could write games in Photoshop, we might get some decent performance out of this thing.") Well, thanks to a combination of Boot Camp and Adobe being slack, the fastest way to run Photoshop on a Mac is by doing it on Windows. No, I know it doesn't mean much, but... the irony, it overpowers!

• The vast majority of the time was spent making differing versions of the final screencast. The two primary reasons for this were:
  1. Practise, demonstration, feedback. This is a good reason. I mocked up early versions of the screencast (usually without speech) and ran them past others (bosses, colleagues, and examples of the target audience) for advice. Given that the screencast was to go on the main site, this was pretty vital. It also meant that some vital things I had forgotten were included, and other pointlessness was excluded - hence cutting the original ten minutes down to a much snappier six.
  2. The ground shifting under me. This is a bad reason. As I say in the screencast, I'm recording it on an internal staging server - a necessity since it was pre-launch. (It would have been nice if the server was able to fake its own URLs for the purposes of the screencast, but I'd already taken up far too much of the dev team's time and they were busy with far more important things, like fixing real bugs. Call this a lack of planning on my part.) Anyway, at least one of the screencasts I'd made was a candidate for final, before realising that what I'd recorded wouldn't actually look like that in the final release of the service. Mark this down to a lack of communication with the rest of the team.
• Coming up with the right dimensions for the movie was tough - I had to make a compromise between the average Ning page size and the available viewport area in a browser on a 1024x768 screen, bearing in mind the height added by the player controls. I settled on 816x550 - narrower than most Ning system pages, but wide enough to fit the important details.

• Most of the initial effort put into recording was made on my office Mac, until I realised that the product I really should have been using - Techsmith's Camtasia - was a far better bet, and Windows-only. Up until then I'd been using Snapz X Pro on the Mac, which recorded into QuickTime's "Animation" codec. The only thing I could find that would edit this without utterly ruining it was SimpleMovieX, a fairly nice bit of shareware only hampered by an occasionally-frustrating interface and a continual habit of crashing immediately after a save. (iMovie and Final Cut Pro were both excluded for their insistence on transcoding to a far fuzzier format.) I'm sure there's some other Mac software that would have done the job better, and if anyone feels like pointing it out, they're welcome. Until then, those recording screencasts of Mac software would be well-advised to use Camtasia with a VNC session.

  Camtasia is a genuinely lovely bit of software, and if you're doing this professionally then it's well worth the $300 price tag. It's an all-in-one recording and editing suite which can publish to Flash, AVI and more. It has a bunch of nifty features I didn't touch, such as "call-outs" (little pop-up bubbles that explain things). It had a couple of irritating little quirks (if you've added a soundtrack in chunks to the video timeline, then want to extend the video early on in the piece, the video gets shunted along but the accompanying audio doesn't) but mostly it was more than adequate and made it very easy to shift, cut and add bits of audio and video until I was satisfied. I'd also tried Macromedia's Captivate, but that's more for taking a bunch of screenshots and animating them in a rather fake (but helpful) way - not so suited for what I had in mind.

  A couple of additional Camtasia-specific notes:

  • For some reason, the default setting for publishing to Flash is to have it wait for the whole movie to load before it starts playing. Sod that. If you're recording at, say, 32kbit/s MP3, the whole bitstream will come in under 128kbit/s, so anyone on broadband will be fine to start playing immediately.
  • I didn't get the point of audio track 1, which won't let you edit nearly as flexibly as track 2. Unless I was missing something, track 1 won't let you easily delete and shuffle audio clips about; track 2 will. So I made sure to always record to track 2.
• I recorded both the audio and video in short cuts of about thirty seconds each - that way, if I got something wrong, I only needed to redo the last thirty seconds. It also made things much easier to chop about.
• The main problem with doing everything in short cuts rather than one long recording is that you have to worry about continuity. It's completely vital to:
  1. Keep the window you're recording in the same size and place through the entire recording session
  2. Finish a shot with the mouse pointer in a location which is easy for you to move back to for the start of the next one - the same applies to page scrolling
  3. When you click a link, don't move the mouse pointer at all until the next page has loaded - this makes it much easier to snip away long loading times
• It's okay to cheat as long as it doesn't affect the main point of the screencast you're making - in this case, that developing new Apps on Ning is speedy, easy and free. The most obvious way I cheated was by magic-pasting the lines of code into the editing area (done with a nifty little system-tray clipboard applet called Ditto). I had to do it this way, because it halted a nasty race between screencast production and my rapidly-disappearing sanity. The problem is that I can't type for toffee, especially not at speed, and almost every single attempt to do so was filled with panic, multiple typos and repeated syntax errors when I hit "View App" only to be greeted with the results of yet another missed comma. I'm sure that others can do it fine, but it put way too much stress into the recording process and ultimately it's not really needed. This is a demonstration of Ning, not my typing. However, I also realise that it distracts from the idea that the performer is creating from scratch, thus dissociating from the essential everyman. Once again, I welcome any discussion more constructive than "haha, you cut and pasted, you LOSER". (Yes, Tim, that means you.)
• That said, there are other speedy behaviours that should be eschewed in order to maintain the focus of the audience. Bear in mind that most of the time the viewer's eyes will be following either (a) the pointer or (b) the text cursor, so jiggling either about like Ray doing the gothic dance is not a great way to maintain attention. (Hence the earlier tips about mouse pointer continuity.) Mouse strokes should be paced and firmly-directed, if that doesn't sound too pretentious. Similarly, sudden use of the scroll wheel can be a bad thing - better to visibly grab and drag the scrollbar. (I wasn't great at this, the main problem being throwing the mouse off the right-hand side every time I wanted to paste something with Ditto.)
• Narration took a few attempts - it wasn't easy for me to speak in a slow and clear way that didn't sound incredibly dry. It took repeated coaxing by Gina before I found something approaching the tone we wanted: Clear, friendly, enthusiastic. (Not saying that I nailed it, just that what I ended up with was a lot better than the start - it's still a bit too fast, and mumbly in places.)

And that's about it. I have a few more screencasts planned to demonstrate other Ning wonderfulnesses, and I'll probably gather a few other eggshells of wisdom for you which I then may update here, if I can be arsed. In the meantime, the last bit of advice I have is: If you've spent so bloody long on a screencast that you want nothing better than to wreak terrible revenge upon it, then you may find it quite satisfying to do so.

UPDATE: Many thanks to Jon Udell for his kind words and additional comment. I feel rather guilty for forgetting to link to Jon in the piece above, since his various articles on the topic were invaluable in the early stages of working out how to do this. I wish I'd been less insistent on trying to do it on my work Mac and used Camtasia from the start, as he advises - it would have saved a whole lot of time. Also, he mentions something else I forgot - Camtasia's nasty habits of outputting 'casts with rather choppier audio than you get in the preview; his advice in dealing with it is excellent, and I hope that Techsmith fix this bug as soon as possible. Thanks, Jon!

Still in the process of blog moving, still about to blog about Ning, etc. But until then, a huge glaring fish in a bucket: Seth Godin's post on legacy reasons. Lots of trackbacks, none of which seem to have caught on that it must be some kind of hoax.

The reason about Blockbuster? Not true. (UPDATE: Still checking this one.)
The reason about places of work? Obviously not true. (UPDATE: Not so sure. I'm thinking about organisational and output-gathering benefits as primary rather than power source, though clearly the power source also has something to do with it. Anyone want to supply evidence of either (which I completely lack, go me)? I'm still interested in this one, though I'm probably completely wrong.)
The reason about typewriter keys? Famously not true. (UPDATE: Boy, was I wrong. And even worse, I didn't read the linked article properly, which clearly agrees with Seth, as Paul points out below.)
The reason about SUVs? True.

So, what's Seth really asking?

(Addendum: I should add that the above question is not one of those A-list style pretending-to-know-the-answer-to-a-rhetorical thing. I have no idea what he's really asking, or even if he knows that half of his reasons are bollocks. But, as Nick has already shown in the comments, it's a fun thing to think about.)

As you can see, I've fixed this blog, only now it's getting hammered with over 200 spams per day ('cos MT-Blacklist appears to be irreperably b0rked). I've installed MT3.2 elsewhere and I'm getting ready to move to it properly, only - how do I clear out all the old crap? There's over a thousand spam messages in my export file, and while 3.2 appears to have all kinds of whizzy barbed-wire anti-spamness for incoming attacks, it seems to have no way of applying said whizziness to the buggers that have already nested. I've hunted through the UI, I've Googled around, I've even posted to the 6A support forums, all to no avail. Anyone? Please? Help?

(And, just to fend off the obvious, let's assume I want to stick with MT for the moment until I'm certain there's no chance of success, okay?)