gurgeous

I certainly was measuring memory consumption. Many of these images are offscreen in paging scroll views (per the blog post), or stuck in caches.

Ooh, cool! We're using JPGs, though. At quality 75:

   67K  flT7c.jpg
  558K  flT7c.png

Once those JPGs make it across the network and arrive on the device, we're stuck with plain old CGImages, right?

I'm not familiar with PVRTC, unfortunately. Here's one of the better articles I could find:

http://www.kihongames.com/blog/labs/2011/04/labs-pvrtc-power...

Some highlights from the article:

- "Avoid using PVRTC when you have static images that are the focus of your content" - "images are 2-4x larger, but memory use is less" - "amazing memory improvement, but no real change in performance" - "textures must be power of two"

It probably doesn't fit my use case, but it's fascinating.

Ha! I really hope none of these techniques are in vogue for iPhone 10. Or iPod 10, for that matter.

I am the author. We don't have comments on the blog, so feel free to chime in here.

He loved the Urbanspoon iPhone app. The iPad app, not so much.

Someday I'll write up the story about the time we met in person. Yes, I too got yelled at by Steve Jobs.

I too had a bad experience with a vacation rental and was unable to post my review to VRBO. That's why I founded Dwellable, a vacation rental site. Here's the full story on GeekWire from a few months ago:

http://www.geekwire.com/2011/hawaiian-beach-vacation-hell-sp...

This behavior on the part of HomeAway (VRBO) is really a natural consequence of their business model. When the property manager is your only customer, it doesn't make sense to post negative reviews.

We're doing things differently at Dwellable, of course.

I participated in the beta. This is a nice app, well done. Dropbox and music - two great tastes that taste great together.

Is it just me, or do they have way too much downtime? It seems like downtime is the rule there, not the exception.

I was in the beta. TouchBase is cool, very handy if you're running around from meeting to meeting all day.

Thanks Nat - I fixed the like button. Turns out that you have to specify the href param when using the iframe, unlike twitter.

I've written a lot of data tools as part of Urbanspoon and subsequent startups. I like to collect publicly available data, clean it up, normalize it, and then release it in a more useful way.

Hot tips for crawling data:

  - Cache pages locally while you work on the indexing
  - Nokogiri is awesome
  - Don't be afraid to use regular expressions
  - Initially, put data into a spreadsheet (not the db).
    That way it can be checked in and diffed.

I also have a lot of subtle tricks for cleaning up messy data. For example, to see if two similar authors refer to the same person, I have a method that converts an author name to an author key. The key is just like the name, only it's been uppercased, apostrophes removed, etc. Plus weird stuff like this:

  # replace all vowels with the letter E
  s = s.gsub(/[AEIOUY]+/, "E")

It's little things like this hack that make a big difference in data quality.

(edit: formatting)

Update - GeekWire picked it up:

http://www.geekwire.com/2011/hunger-science-fiction-books-sp...

Also, thanks to webwright for the title suggestion.

Here we have something called the Standard Health Questionnaire (SHQ), which they make you fill out when applying for insurance. But there are a bunch of exceptions at the top which make it possible to avoid submitting the SHQ. Running out of COBRA is one of those exceptions.

My family rates assume one 10 year old child and spouses that are the same age. Could that be the problem? I tried to add disclaimers to that effect. Maybe I need to make them more prominent. Certainly the rates match up quite well during my testing.

Twitter Bootstrap - highly recommended!

I've toyed with adding some elaborate calculators. Last time I went through this exercise I built a rube-goldberg spreadsheet that let me plug in doctor's visits, major illness, deductible, premium, etc. and spat out a yearly number.

It was just too complicated, though. Most people don't even understand what "coinsurance" is. I might add some pretty calculators later if the site gets some traffic.

Not a bug, sadly. That's just the way it is in some spots. We're lucky up here in Seattle.

Fixed, thanks! jQuery Tablesort has a bug with currencies and commas...

Twitter Bootstrap is completely awesome. I could never make it look this good on my own, especially with full browser testing.

I'm trying not to clutter the UI. Ethan (my co-founder from Urbanspoon) came up with the idea for the "office visit" icon. I used to have that in a column and it was just too much, even on my enormous monitor.

More like kayak? I thought about adding the sliders, but it feels like the filtering is sufficient.

I've been in your shoes and I empathize. In fact, I was in your shoes just a few days ago when I applied for insurance. The rates here are "representative" and I tried to add disclaimers in the obvious places. The rates are similar to what you see if you check with your state's insurance commissioner. Most states require the health insurance companies to publicly disclose those rates.

Of course, they can still mess with you and they won't hesitate to do so if they think they can't make money at this price. There's no way to find out the final rate without completing the entire application process. In that respect it's similar to auto insurance.

PickHealthInsurance still has a lot of utility compared to existing sites, though.

You can get often get the advertised rates, but not always. In Washington State for example, you can get these rates if you're running out of COBRA and you've been covered continuously. They aren't allowed to ask questions.

I don't think there's a magic bullet, unfortunately. Suggestions welcome.

I'm releasing the first version of PickHealthInsurance today. It helps you compare health insurance plans (individual, not group) in the US.

I put this together over a few weekends. My COBRA is about to run out and I found it exceptionally difficult to compare plans and prices with the existing sites. This is the third time I've had to buy individual insurance and I finally decided to do something about it.

Things crystallized for me when I almost bought a plan that claimed to cover maternity, only to discover late in the game that it had a separate $20,000 maternity deductible! What a mess!

I have a few goals with PickHealthInsurance:

- start showing approximate rates almost immediately, don't pester me with a bunch of questions

- explain confusing terms like "coinsurance" and point out the difference between a PPO and a POS.

- show plan stats up front, not buried deep within the bowels of the application process

- make it easy to use and blazingly fast

Built with:

- Rails 3.1 (HAML, Sass, Coffeescript)

- Twitter Bootstrap CSS

- Deployed on Heroku/MongoHQ

What do you think?

I am the author. Teleport sets up servers by "teleporting" to the machine and running itself there. It sets up Ruby, then follows the instructions laid out in your Telfile to customize the server. Here's a sample Telfile:

  user "admin"
  ruby "1.9.2"
  apt "deb http://downloads-distro.mongodb.org/repo/ubuntu-upstart dist 10gen", :key => "7F0CAB10"    
  role :app, :packages => [:memcached]
  role :db, :packages => [:mongodb-10gen]
  server "server_app1", :role => :app
  server "server_db1", :role => :db    
  packages [:atop, :emacs, :gcc]

More details on github. Feedback welcome!

Money quote based on trying to restore their MongoDB during the EBS downtime:

"Waiting for indexes to build takes approximately as long as it takes AWS engineers to fix EBS issues."

That can happen if we can't find good overlap between your work history and the work history of your LinkedIn connections. We make reasonable attempts to normalize the resumes, but we can't account for variations, or even plain ol' typos. Double check your work histories and see if you can line them up better.

TechCrunch drives a bunch of traffic, generally, especially something like Cubeduel that plays well to that crowd. We're somewhat insulated in that many people lack LinkedIn accounts and bounce off the front door.

The first system to fall over was our message-queue that talks to LinkedIn for logins and refreshes. I had to spin up a separate machine at Rackspace just to help logins along. Hopefully that'll hold things steady for a bit.

We don't have those kind of detailed stats yet, unfortunately. Right now we're just trying to keep things humming along. It's just me on the tech side!

Resto duel is an interesting idea. Ethan (from Urbanspoon) actually mocked this up at one point but we thought it was kinda lame. The results just weren't that compelling, sadly.

Adam Doppelt (co-creator) here. Feel free to chime in with technical questions and I will attempt to answer them. That's half the fun of a side project!