(this is page 116 of 121)





Latest

So I am back from my holiday experiments with Squeak and Seaside. I guess it's called experimentation because you are supposed to play without necessarily having a real agenda. This is a good thing because despite the fact that Smalltalk is a language I've been wanting to learn for a long time, and despite the fact that Alan Kay is one of my all-time computer heros (along with Ken Thompson and Dennis Ritchie and Knuth), I was sort of left ho-hum about the whole thing.

First, Squeak: while the idea of a fully integrated environment in a great thing in principle, Squeak is just too baroque and isolated from the surrounding environment (the web, the OS, etc.) for me to really envision using it as my primary development environment. From the trivial (the fonts supplied with it are just a complete eyesore) to the significant (in full screen mode on my 23 inch display the UI does not redraw correctly or respond to mouseclicks in a consistent way), the whole environment feels stuck in 1980. It may be better on Windows (I used in on a 2x2GHZ G5) but running inside a window in OSX made it's retro feel painfully obvious. The Squeak browser pales in comparison to Firefox and the web and basic OS-integration is sorely missing (sort of like Java apps from 3 years ago). The class browser looked like an awesome tool but too many things about the way you could scroll the panes was just broken in my view. And finally in some cases, like when grepping for a method name with the "method finder" the whole environment would just freeze up for 10-15 seconds at a time becoming completely unresponsive.

Then there was Seaside: Very cool idea, and probably a great implementation, but so sorely missing in documentation that it's pretty hard to go beyond the basic trivial counter application. I was particularly interested in how the continuations-based pattern for web development dealt with persistence (beyond with keeping things in RAM on the image) and concurrency (users working with the same model object at the same time) but I found little help in either case and it was not immediately clear to me from browsing the class hierarchy how this was supposed to get done.

The one thing that Seaside did have which is worth copying was a really neat way to inspect all of the page fragments (or widgets) from the web browser (outside of Squeak) including a fully web-based class browser. This was pretty awesome and it's a testament to Seaside (and a discredit to Squeak) that I found it better to work from the browser-based interface than in the cumbersome class browser-transcript-debugger inside the Squeak environment.

I still think that the Ruby-on-Rails podcast on Seaside with Avi is well worth listening to but I am not sure how he manages to be so productive with a tool that, at least after about 5 hours with it, seems so much like a university research project, and not a full-blown development environment. Maybe I lack the vision for it, but for now, I'm staying above the waterline for real work.

Comments Permalink

Down By the Seaside

A must-listen from the Ruby-on-Rails podcast (has very little to do with Ruby on Rails): an interview with Avi Bryant on a new Smalltalk-based web framework that breaks the straw-thin mode of thinking about programming for the web (session-less back-and-forth of key-value pairs does not have to be the state of the art).

This is one of those interviews that makes you wish podcasts had a way to "call in." There were just so many more questions that the interviewer should have asked but didn't like: how you deploy faceless Squeak-based applications? How you distribute them? And why is Smalltalk really is so much more efficient than Ruby/Python/Perl?

For everyone with some downtime over the next few days, spend some time down by the Seaside and think about a better world.

Comments Permalink

Interface, Learn Me a Lesson

A recent Tech Crunch piece on a new photo sharing service called BubbleShare has got everyone excited about the power of simplicity. Bubbleshare amounts to a very thin (mostly email-based) interface for gathering collections of photos as "albums," along with a very nice set of DHTML effects (all built of the marvelous scriptaculous Javascript library).

Here is the good: BS has a wonderful out-of-the-box upload experience. You can get your stuff up there without a laborious account signup process that requires you to enter in every last bit of info only to then wait for an email with a too-long-for-its-own-good confirmation URL that you have to click on to get started. The BS team clearly worked hard to get this part right and it really shows.

Here is the bad: this is not an application but an applet, or perhaps better yet an tinyplet. In BS I don't have a notion of a library except for a bunch of emails in my inbox with byzantine Dewey Decimal style URLs that are also too long for their own good. There is no notion of permissions and sharing is limited to "forward this cool link to your friends" which, while great for videos of martial arts mishaps, doesn't map to the semi-private way in which most folks want to share photos (and security through obscurity can only take you so far).

And worst of all, despite the nice page title "Tell your story", BS provides very little in the way of tools to really tell stories. Where are the rich templates that incorporate text as a first-class citizen? Where are the picture editing (think crop not Photoshop) tools that really allow you to leverage the power of images? And where are all of the cool ways to tell stories online that folks have been dreaming up for some time now? And most importantly, where is the community?

Having dug around the site a bit, I realize that the BS folks were working with a role model in mind:

When we set out to build BubbleShare, the first question we asked ourselves that I can remember was: “If Google created a photo sharing service, what would it look like?”

But Google is mostly search, not expressive tools for personal publishing. To me this is like saying: if Cisco built a photo-sharing site, what would it look like? And you can bet that if Blogger (which is Google after all) had set out to build a photo sharing service it would most certainly not look like this. Because in BS there is precious little opportunity for context on the part of the story-teller and blogging in general is all about story-telling in context. Whereas I might spend six hours tweaking my Wordpress template, in BS there is no chance for a creative investment.

There is this general trend with all of these Web 2.0 companies to minimize the interface to the point where we're back to 1995 CGIs with two form elements and nothing else (witness the recent Odeo makeover). While I am a big fan of a lot of the tenets of Web 2.0, this is not one of them. Instead, I'd like my webapps to provide me with an interface that I can learn. Sure, let me do the basics fast and make it dead-dumb simple (as someone from the PC era once told me, the key is to get the user to a moment of absolute pleasure within 5 minutes). But then give me enough depth to sink my teeth in, to learn by struggling with the app, and to be rewarded when after struggling my creative investment pays off in something that really lets me express myself.

We're not dummies that either get a blank page with two form elements or click away to the latest Hollywood drivel. We want to be challenged, and if we're going to switch photo sharing sites, or start using them over email attachments at all, it better be because they really make something possible which just wasn't before.

Comments Permalink

It's the Product Stupid

The number one justification that I hear these days about why so-called Web 2.0 companies succeed in areas where better-funded Web 1.0 companies failed is the "maturity" of the online audience. Ofoto had to dump millions into marketing online photo sharing because people didn't understand what you could do with digital cameras and the network. At abuzz, we were too early for people to understand that the site was really a mix of the best parts of Friendster, Craig's List, and the current crop of Q&A sites.

Baloney. This has always felt like the easy way out to me. It would be one thing if the "before" world consisted of people with 14.4Kbps modems and gopher file requests, but the reality was that the web of 1998 looked a lot like the web of 2005 (the penetration of broadband being the main material difference). We may have had more browser hoops to jump through in order to get decent usability but people were still able to grok the power of the network and the implications of communities online even back then.

Which is why it's so refreshing in the wake of the Yahoo-Delicious deal to read this piece by Ari Paparo on why Blink failed with more money and time where Delicious won. In the final analysis Paparo claims that it always comes right down to product:

My intent was to show that product design matters. We had more money, more users, a five year head start, and some really, really smart people working on bookmarking in 1999. The bottom line is that we simply didn’t get it right.

I'm ready to admit that on the web it's all about getting the product right. This is exacerbated by what someone yesterday described it to me as the "hit-and-run" mentality-- since it seems that just about every day someone is announcing a new online service, a key segment of your early adopters are likely to "hit" your site, make up their minds in 5 minutes (if you are lucky), and quickly run to the next thing if you fail to capture their hearts and minds.

Comments Permalink

Adding to the A-list

I've written before about how I feel about great people and what they do when put together under the right circumstances. Well today at "Secret Startup X" (name to be announced before the end of the year), we picked up another great person who is coming on to help change the way people tell stories online.

And yes (to all of the Django people who have asked), Ned is going to be working with Django. We've got a sort of layer cake approach to what we're doing that picks the right tool for each task. We're doing a lot of stuff with images (storing and transforming them) so we've got a bunch of C. We're doing a fair amount of AJAX work so obviously we've got a load of Javascript. And in the middle of that great big Oreo cookie is a whole load of Python running inside of Django.

Django is great for a lot of reasons (so many that they deserve their own blog post). It does just enough without trying to do too much magic. It's relatively bug-free. It's surprisingly snappy. And you don't have to "buy in" wholesale or risk feeling like you are fighting the framework all the way to your task.

Anyway, the team is super excited to be getting another castaway on to the island. And we are all eagerly waiting to show him that-- at least in this category-- software doesn't have to suck.

Comments Permalink