bjclark

To elaborate on qhoxie's statement, NoSQL is great for some things, but there is this idea out there that it's a panacea for scaling your webapp. And while all the things highlighted in the article are performant, few of them even scale, and none of them are perfect.

It's not a big deal when you do it the first time. It's a big deal when you need to start adding servers. What do you do?

I'm currently in an engineering role at a startup, but I have a BFA in Fine Art, do design and user experience design.

Great tips. I've built a couple e-commerce sites for clients over the years and all of these ring true to me.

I think the "Quality Photos" tip is the best. For certain products, allowing customers to submit photos is also great (think Threadless.com style) since it's both product shots and social proof.

Architect

I highly recommend against glusterfs. We used to use it at my current employer and had nothing but trouble serving large files from it. It was fine until we starting putting large files in it, then healing would randomly bring the whole server to it's knees.

Yes, all rich people are alcoholics with trophy wives. You've nailed it.

Don't think of it as starting with the "usability" part. This is starting with the "user" part. What do your users want? What do they need? Choosing the users that you want to support and what features they need will drive what you need the backend to do and thus the underlying data-structures needed.

Is a comment here a Comment on Comments on Comments on Zeldman's XHTML WTF?

I've posted a reply to you on my blog (cause it's too long for here): http://bjclark.me/2009/07/03/how-do-you-redesign-something/

Excerpt: The biggest thing to manage in a redesign is risk. Risk you alienate your user base or do something that negatively impacts revenue, or whatever you're most afraid of.

The easiest way to manage it is to shorten your feedback cycle (not user feedback, any type of feedback) to the smallest possible interval. When talking about design, this means you need to spend the least amount of time on the design you can before you put it in front of someone that exemplifies your average user (persona).

This is a great post. I think this new UCD focus for Agile is the best thing to happen to it since it's widespread adoption. If you're interested in this topic, Jeff Patton's blog is awesome too: http://agileproductdesign.com/blog/

One thing that I think 37s is great about (and I don't know this as a fact, but from the outside it seems this way) is that they definitely trust the people they hire to work for them. I'm not sure that many places would hire someone, and then as them to re-design their account screen in their main application in the first week. And trust them to do a good job!

It's free? What a deal!

getsatisfaction

Hive is great, but it's noted that like most Hadoop things, it's alot better when you have 100 machines than when you have like, 2.

There's nothing stopping you from just running Map/Reduce scripts. Hive just compiles the SQL down to Map/Reduce.

We just launched a widget too and were surprised by how hard it was to stop the cascade. This is pretty useful.

I think you should launch as soon as your service provides enough value to someone that they should care. If no one cares, it doesn't provide enough value (for whatever reason). If people do care, then you know your on to something.

If you launch before you provide enough value, no one will care.

If you wait to launch, you are only doing yourself a disservice by delaying or prolonging the feedback cycle. You should be doing the opposite, figuring out how to make your feedback cycle as short as possible.

Worrying about competitors at this stage is dumb. You could only be so lucky as to have an idea good enough that someone would care enough to try to challenge you for it.

We use GlusterFS and hate it. It's healing seems to do more bad than good and we're looking at moving all of our files to S3 instead.

I'm liking RubyMine and 1.1 has support for Cucumber. It's not as polished as Textmate, but it has light years more functionality. Split window, basic refactoring, search that doesn't blow, etc.

Yup, Ward is my boss. He joined AboutUs over 2 years ago.

Portland is a great startup town. There's a few VCs specializing in PDX startups (Voyager Capital from Seattle is one, they funded AboutUs in Dec).

And we hired another news.yc member qhoxie(http://news.ycombinator.com/user?id=qhoxie) from here to move to Portland and join AboutUs.

I'm famous on Twitter.

This is the most inane headline for the topic.

A. Journalism isn't going to die. Newspapers might, but Newspapers != Journalism B. They don't need hackers, they need entrepreneurs. As cool as hyperlocal news is, it's mostly just an iteration on the newspaper. It's not changing the game in any way.

The only thing I see wrong with newspapers is that the medium doesn't sell advertising very well. It only worked for so long because it had all of the market. Where else would you advertise in the 50's or 60's? I don't buy the "Yesterday's news" argument. I don't think that matters to 99% of the people. If that really mattered, TV would have killed off the Newspaper with CNN. Hell, radio would have killed off the newspaper if it mattered.

I would never be stopped for applying to a job because it's "Senior". If you know your shit, it doesn't matter how old you are.

Well, it's an implementation of the original wiki, but not the same exact implementation the original runs.

I think you're missing the point. You can't bash Ruby because you wrote shitty code in Ruby. What do you want to bet I can write a shitty message queue in Scala? Does that mean Scala is bad? "Reliable, high performance, correct and maintainable" have nothing to do with the language you've chosen.

You dismiss anyone who's job applies to the topic at hand? Shouldn't you dismiss Alex because he's writing a book for O'Reilly on Scala? Should the only people allowed to discuss this be the Erlang people? Just asking because I'd bet they would say that Kestrel is a crappy version of RabbitMQ.

I think this advice matches up really well to one of the things you say in your "5 Founders" essay about Paul Buchheit:

"PB made a point in a talk once that I now mention to every startup we fund: that it's better, initially, to make a small number of users really love you than a large number kind of like you."

I can't say enough things about my Mirra. I also agree that it's 50% more comfortable than an Aeron. I think they have much better back support.

Very timely. I'm presenting at Rails Conf this year and the last time I gave a presentation was, uh, high school. :\