I tried to show how the onslaught of visitors from my Hacker News link last week had an effect on both the server (as shown in all of the 7 day munin graphs), and the project (how I'm responding to feedback and planning the course of NewsBlur for the next few months).
I'm just so close to NewsBlur being profitable. I am 35 premium user accounts away (for a total of 80 accounts). Once NewsBlur goes profitable, that's when the real hard work starts. I would then have support issues, broken features in edge case browser/OS configs, and the headaches of keeping a server running for high availabaility, while on a shoe string budget and working super-part-time. (I write most of NewsBlur on the A train, 35 minutes each way from Brooklyn to midtown).
Wait! How the hell do you code in the A train? Actually this is worthy of a blog post - "How to code during the NYC subway commute". But seriously, I am curious what you do when there is no place to sit or when u r tired (its easy to fall asleep in trains)or what kind of laptop you use!
I have a 15" MacBook Pro. It's the width of my legs.
The hardest part about coding on the A train is when dancers/singers come on. I love watching performance theatre, especially on a train. I just stop coding, put the lid down, and enjoy the show. And if you know the A train, you know this happens once a week.
Otherwise, I am just so in love with coding NewsBlur that I can focus. I talk to myself a little (quietly), drawing in the air with my finger, and am just a little NYC crazy that nobody questions me.
I get a seat 19/20 times. (That's once per two weeks that I have no place to sit). I bring a small book, just in case, so that I at least have something to do on the train.
I've missed my stop a few times. It happens, but it gives me an extra 5 minutes to work, so I don't sweat it.
Hmm, coding on the subway as a blog post. Brilliant! Ok, next week it is.
Somebody asked what I am using to generate those server graphs. The answer is munin. I love munin so much. I use it to monitor both the server and any custom numbers I have.
Thanks for the postmortem, it was very inspiring (especially the image in my head of someone hacking on the subway)! The idea of an intelligent aggregator is something that I have thought about often so I find the idea of NewsBlur to be very interesting. I'm going to check it out.
I got the idea from Paul Ford, also known as @ftrain. He wrote a substantial piece of work entirely on the F train, which makes me think he lives in Park Slope, Brooklyn, and commutes to somewhere along 6th Ave in Manhattan.
I take the A train because it's closest to my apartment, but also because of Ella Fitzgerald singing "Take the A train." Now, if you're form NYC, you'll know that the A train is an express train, so when I see the C train, which is a local train, I'll take it to squeeze 5 more minutes out of my commute. The other benefit is that when folks see both the A and the C pull into the station, they all run to the express train, leaving me a greater chance for a free seat on the local train.
(I write most of NewsBlur on the A train, 35 minutes each way from Brooklyn to midtown)
Maybe its just me, but, there's something nice about working on a train. There's no wifi (at least, on trains run by the MTA) and the train sounds make for nice white noise.
I used to have a ride on the LIRR from Nassau into Penn for an internship. I'd bring my laptop with me and get a good hour's worth of coding, per direction. At least, when I got a seat.
I am unclear on how 80*12 dollars is profitable, altho, I guess if you only worked on this 70 mins a day for a year, I guess that comes out to 4 dollars an hour, but you must have worked on this on the weekends as well...
I guess you are only factoring the cost of servers, not your own time.
I'm impressed you could do this sort of work on a train. In trains/airplanes my working capacity seems to be limited to emails, reading, and brainstorming. Sketching designs is the closest I can get to coding.
This is a product space/market that many have attempted in the past and failed. I have probably used a dozen different feed readers in the past 10+ years, and this is the first time that I have signed up for a new product in this space and been genuinely excited.
Great work. Content publishers will also like this because you aren't scraping, and you are preserving the original format of their content.
(Bug report - are you auto-detecting feed urls? I put in the domain for my own site and it didn't find the feed, although it is referenced as a rel. You also don't have Sydney as a timezone)
If he's using Python, he's likely using "feedfinder.py", by Aaron Swartz. It doesn't find your feed at all.
It looks like your feed isn't being served with an appropriate mime-type -- or at least not one that it understands. Seems to be failing in the 'test' method.
Never heard of anyone using this content type. It's typically the application name with XML as a modifier, "application/xhtml+xml" or "application/atom+xml". "text/xml" is the generic catch-all but is not valid for feeds.
This is brilliant, I'm switching over to this right now.
Funny enough, Google reader has just died a quick death since I discovered HN a month ago, but maybe I'll start using RSS again with News Blur.
Obviously this is beginning stages, but i think the typography and sizing of UI can still be refined and reduced substantially.
The intro page is the weakest for me. The calligraphic doodle below google reader feels alien. RSS reading is all about streamlining your reading. I'd dull down the colours of the UI to the max (think the new itunes monotone scheme) in order to maximise the individuality of the fed content (which is the USP of your product).
You only need The Google Reader Sign Up and the How it works shown for the beginning. The New Improvements feed is not what a first time viewer needs to see when they put thier virgin eyes on your product.
Not to turn this into NewsBlur tech support, but is there some way to see recent unread posts from all my feeds, in chronological order, without having to click around on the left sidebar?
That is basically my single mode of operation in Google Reader. I'm sort of surprised that I can't do it trivially because I thought that everyone used RSS readers that way.
It's river of news and it's slated for next month. I wasn't going to implement this until recently, when a user very persuasively convinced me to make this a priority. I thought it was a workflow that didn't fit in with the rest of NewsBlur. He thought it was a necessary feature for any hard-core user.
Everybody reads RSS differently, and I am only trying to pleasure a few groups for now. But river of news is one of the highest ticket priorities. Up there with the iPhone app (which is 60% done), fixing misbehaving feeds, and fixing the many Chrome bugs.
Yeah, I was figuring out how to do this as it's the way I read everything (Twitter, Google Reader, Facebook).. rather than visit sources one by one (which is what I'm trying to move away from for sites I like the design of - if NewsBlur can combine that "site's own pages" effect with the river benefit, a big win for me).
One thing I can't figure out, though, is why many feeds do not show the "real" page for the item. Instead, it reverts back to just using the feed or showing the front page of the blog in question, even though the URL is clearly in there. When I'm on "Original" I'd just like to see the URL for the item on the right without having to open it in a new tab/window. Or is this a bug?
It's a different use case. If you want, you can click on links in the original site, but then you are no longer on newsblur.com. The way i am able to take over the original site is through a proxy.
What your describing is that every link shows it's original site endpoint, which means reading every story would require a separate page load. Instead, I just consolidated the front page, and if you want the story content as it appears in the feed, then there's the Feed view (and you can go back and forth between them with the left and right arrow keys).
Otherwise, it's a slow process to load all of those stories one-by-one.
Aha, I thought it was just an IFRAME (and then thought the "arrow" at the side was rather clever and I'd need to go see how you pulled that off ;-)).
Oh well, that's a shame. One of the initial big wins (in my head) was that I like to see the original design of the item rather than the plain feed version (handy for these individually designed blog posts nowadays) but without opening 101 tabs over my session. This still works fine in NB for blogs where all the posts are held on the front page a while, it seems, but just not those that nest things on individual post pages I guess. (Hurrah for you open sourcing it though!)
It's refreshing to read an article all about the service you're providing to your customers and not your percent conversions or CPM. So many of the entrepreneurs here are all about the money rather than the product.
Hey just one small but really noticeable improvement:
When you change feed, and the "iframe"(or whatever it is) is loading the next page, you should fadeout the old one, and maybe (if possible) put a loading gif or similar until load completion.
That would make the interface far more responsive, since when i was clicking feeds to test, sometimes i didnt notice if the click was doing something or not, and that produces frustration along with multiple clicks (unnecesary work overload)
I really hope, and believe, this will replace my use of google reader, which I have found to be absolutely painful to use. I wish google products had more inspiring interfaces.
I'd love to talk to you about your decision to open-source it sometime.
Yes, that whitespace is intentional. I know, it's a bit disconcerting because nobodyd oes it that way. I may fill it with dashboard style analytics and graphs. But that's a ways away for now.
The focus is on the sites. The logo only shows up when you first go to newsblur.com. That's also intentional. Why do you need to see the logo as you're reading your sites?
As for why I went open-source, I figure that the good will that comes from being open-source will help me obtain my #1 top priority: meeting future co-founders. NewsBlur is my public resume. Making it profitable is priority #2. Well, it can easily be profitable if folks are buying premium accounts and not bothering to host everything on their own servers. Seems to be working out pretty well, so far.
This is a cool project and really great that it's open source, since I want to learn more about Django, but that whitespace really stopped me with I first loaded the site. I spent several minutes thinking it must be a CSS error. It made me wonder if there was a layout problem, and how did the developer miss this. So maybe it's at least visually distracting, which takes focus away from all the good work done on the site.
Hah, thanks. She's actually blind, hence why her eyes are bigger than normal. But she's a beagle, so if you know beagles, you know that they couldn't care less about their eyes. It's all nose for them. On walks I have to remind her that she's about to bump right into a tree.
And that truck for an icon? It's for all the food I have to truck in to feed her. Har har.
I've been spending too much time on HN and not enough on RSS. Which captures the most interesting high-leve tech stuff, but doesn't keep my in all the loops I want to be in, and I have Google Reader fatigue, so I think I'll give this a serious look.
Hey Samuel, I met you at the Hackers and Founders meet up in NY a week ago. I remember you telling me about Newsblur and so I took a look at it and I am very impressed and will continue to use it.
Yes, I did make the UI. Uhh, as for graphics, I use Silk icons. Besides that, I don't think I use any graphics. I'm very big on thinking about design, but I do not for the life of me know how to draw.
I consider my work as the epitome of the "Design is not just the look" ethos. I spend what little free time I do have thinking about the flow for all the little parts of NewsBlur.
I'm just so close to NewsBlur being profitable. I am 35 premium user accounts away (for a total of 80 accounts). Once NewsBlur goes profitable, that's when the real hard work starts. I would then have support issues, broken features in edge case browser/OS configs, and the headaches of keeping a server running for high availabaility, while on a shoe string budget and working super-part-time. (I write most of NewsBlur on the A train, 35 minutes each way from Brooklyn to midtown).