How Pocket Hit 20M Users with 20 People
firstround.com
firstround.com
Alas, their quality has been going down recently. The most important feature - keeping the location inside an article between app invocations, is not working. For long articles, if I stop in the middle and want to resume later, there's a 80% chance that Pocket will happily set me right in the beginning.
Besides, their rendering for articles with code sucks, so I almost always use "web view", a decision Pocket also forgets every other time.
So I end up using Pocket as a convenient keyboard-shortcut to save articles, but on my phone actually open them into Chrome, which has no problem remembering the location in a tab.
How do you handle security. By allowing anybody to push arbitrary links/text to your phone aren't you opening a vector of attack?
I will fix the link. Thanks for point out.
I worked for MultiScope in 1991. We had to order discs and have disc labels printed, copy the compiler onto the discs, have manuals and boxes printed, stuff and shrink-wrap the boxes, ship to Ingram Micro for distribution, and then wait 2-4 weeks for our product to show up on the shelves at Egghead. I recall 5 developers, and we were ecstatic to ship 4,000 copies of a major new version.
That gets me thinking in terms of leverage. The leverage that 2015 Internet technology affords a single developer is a potent economic force.
Late 80s/early 90: the typical software product was shrinkwrapped, cost $500, sold in the thousands to tens of thousands.
Now: "Free", ad-based in some form. A few dollars per user at most. More users.
We need an ongoing list of startups and their spirit animals. How else will I know what products to use.
But I signed up for Pocket because it does two things very well. It lets me send things to my Kobo e-reader via email and gets around pay-walled content. I don't recall ever seeing ads either.
I don't pay for the service because I don't need the for-pay features, but would in fact pay for the basic service if it wasn't free. I wouldn't even mind if they in turn payed NYT, New Yorker, etc for the content I'm getting.
Is it a copyright violation to make a cross-stitch version of a tweet for your living room? To provide a meme generator service that uses NYT headlines?
Here in the U.S. this kind of copying by end users is probably fair use, which is a defense to copyright infringement. There is a line of cases that authorize home recording of video broadcasts. I think the reasoning upholding this "time shifting" or "format shifting" would also apply to a read it later service.
It is very unlikely that a publisher would ever try to sue an individual reader for using one of these services. It is more likely they would try to shut down the service itself by claiming that the service induces or is liable for contributing to the end-users' copyright infringement. But these claims would fail if what the users are doing is fair use.
The weakness of the U.S.-style fair use concept is that it is a case-by-case determination based on several vague factors. Getting a legal case to the point where this determination is actually made is very expensive, so we don't have much precedent to rely on for new kinds of activities. And what precedent we do have is not always easily extended to new activities or technologies.
There's a nice overview of fair use here if you're curious: http://fairuse.stanford.edu/overview/fair-use/what-is-fair-u...
With a customer facing transaction based service like Uber this would be hard. But pocket is pretty straightforward.
Also it's not really easy to prevent against such security things with small teams if you can't rely on the tooling inside your language. Building Crypto things is really hard. So People needing to rely on. Today I've rewritten a authentication / authorization layer for one of our internal Apps, now the App is hardended against csrf, timing attacks and many more things. The only things that I used was the internal PBKDF2HmacSha256 for Passwords of Java Sha256Pnrg for Salts and I copied behaviors of the Auth0 JWT Library (i just used another json implementation which the app already had) and i have something like 200 Lines of Code. It's secure as long as MessageDigest and the PBKDF2 Code of Java is secure and that's something people can't avoid. Especially in small teams.
Nowadays though, as long as you have a good idea and some reasonably basic skills, you can make an app for millions of people.
[1] https://getpocket.com/blog/2015/06/pocket-is-now-built-into-...
> How many of those users were forced by updating Firefox?
"forced" would be zero
I bet the number of users would be much smaller if there was no such button.
How hard is that to understand?
> Force [verb]: make (someone) do something against their will.
1. has a decent HTML parser that doesn't omit images from articles all the bloody time
2. implements a proper state restoration mechanism for its iOS app. It's exhausting having to scroll through my entire article list trying to find the one I was reading (and then finding the original position) every time I switch to Safari to look something up and then back
3. has a highlights & notes feature as a bonus
Please?
Pocket's inability to remember the current article/position is one of the most obvious, annoying, and (at least by me) reported bugs ever. It makes the app close to useless. Yet they're unwilling to do anything about it. So I'll happily switch for this alone. And highlights/notes is a dream comes true
Going to give Instapaper a shot. Thanks!
Back in the day when we built the Trapster.com website (prior to the Nokia team takeover) we a/b tested a static user # that we updated monthly vs. a # that changed and was specific to usage / etc.
We found that the specific was a significant conversion driver and we executed several tests to get that message closer to the top (2).
Of course this was also back in the day where consumers looked on desktops for the app to download vs. the modern trained user that looks on app store.
But maybe something to consider moving up the page and building a script to execute more frequently.
(1) http://web.archive.org/web/20150325061108/https://getpocket....
(2) http://web.archive.org/web/20130812132821/http://www.trapste...
I have a Kobo ereader and so use Pocket for saving articles that I want to read later, away from the computer. Have found it works very well for this. (That said, if my ereader integrated Instapaper or Readability I’d probably be just as happy.)
My one complaint is that every now and again it finds an article it cannot extract the main content from. In that case it never ends up on my ereader, and there’s no obvious indication that there is a problem.
To track my readying habits, I wrote a little PHP browser based application that interfaces with the Pocket API (and the hn.algolia.com API). Once I’ve read an article I archive it. Then when I’m back at my computer I run my app, which lists the archived articles, any related Hacker News pages, and lets me manage the articles (delete, save locally, etc.). It makes it easy for me to follow up and read the HN discussions after I’ve found the time to read the article.
Naturally I called it Pocket Lint.
As for the Web Clipper, I don't think that there's any similarity.