I’ve been interested recently in Android Market; it, and its peers such as the Apple App Store and Ovi and (in a slightly different flavor) the Google Apps Marketplace are all recent arrivals, trying to do a new thing. And I don’t think any of them are doing it very well.
(While I work organizationally right next door to the Market people, at the moment I’m not really close to their internal architecture or long-range plans; I still have something of a black-box view.)
Right now the consensus seems to be that Apple’s is the best among the app-retailing alternatives. But none of them are actually very good. In terms of sorting the wheat from the chaff, and helping you find what you need, these services fall way short of what Amazon provides for media and what Google provides for the Web in general.
Quick example: I searched for “Vancouver Travel” on the App Store and got a pretty-uninspiring overview page with six iPhone apps (no indication of how they were chosen); clicking on “See All” yielded a horrid 78-app splodge, with no obvious organizing principle that I could see. Let’s be fair: the same search in Android Market yields exactly two results; so I’m not claiming the alternatives are better, I’m just saying that I’m seeing evidence of a hard problem that hasn’t been well-solved yet.
The Firehose · I sincerely believe that it’s hard. As I said, I’m not a Market insider yet, but I’ve spent some time looking at the raw firehose of incoming apps. It’s mind-boggling, overwhelming, terrifying.
The Market is growing furiously just at the moment, probably under the influence of the arrival of lots of interesting new devices. Just as with the Internet itself, Sturgeon’s Law applies: almost everything is crap.
Very high-volume crap. Coming, in most cases, from developers nobody’s heard of, so branding doesn’t help much. Likely, in most case, to sink without a trace, which is a real tragedy because there are nuggets of gold among all the dross, things that would enrich buyers’ lives and developers’ wallets if they could only get noticed.
I look at this torrent and try to think of how I’d go about making it more useful and yeah, I have ideas, but man, it’s tough, and I’m not expecting miracles any time soon.
Amazon’s Advantage · Even though Amazon is selling way more titles than any app retailer, the problem they’re solving is more tractable because there’s a whole ecosystem, which includes a big chunk of the world’s academic community, devoted to discovery and criticism of books and music and movies. It operates against a fixed background context, rich with powerful brands; examples include J.K. Rowling and Madonna and Johnny Depp. As we gaze across Amazon’s nearly-infinite virtual retail space, we’re all standing on a platform of largely-shared perceptions of what it is we’re looking at, and for.
The app ecosystem just doesn’t have that. It’s being made up as we go along. Such powerful brands as exist have been imported from other media: Marvel Comics, the BBC, eBay, Monopoly.
Can something as useful as Amazon be built without a vast substrate of cultural commonality? I hope so.
The Search Engine Advantage · Google and its competitors are astoundingly effective at taking a vaguely-worded request — for example vancouver travel — and returning results that are rich in structure and supporting apparatus, not to mention (in that example at least) highly relevant commercial pitches.
This works partly because the Web’s resources are richer in content and metadata than your typical online app listing. But even more important is the rich network of links that serve as grist for the search algorithms’ mill.
I suspect there’s a lot of potential there. Certainly, the retailers could do a better job of making their offerings first-class citizens of the Web by giving them easily-discoverable URIs and encouraging direct links. But anyone who thinks that any existing search engine’s core algorithms can be replicated cookie-cutter style for app retailing is in for a disappointment.
Here’s an interesting offering that crossed my radar the other day: appsfire: “Discover & share the greatest apps”. The fact that it exists is an indictment. I haven’t the vaguest notion whether this kind of approach will work, but I’m glad someone’s trying it.
It’s a Mess · None of the online stores are much like any others, which is a user-experience problem. There’s a huge first-mover advantage, because the first competent offering in any particular niche will get the big download counts and the masses of reviews, creating a positive-feedback loop that makes it really tough for latecomers.
And there is a painful, obvious lack of quality filters. I don’t believe for a moment that the front-end-loaded Apple App Store approval process is anything like the right answer, but as of now I don’t know what the right answer is.
It’s sure an interesting problem.
Contributions
Comment feed for ongoing:
From: Ben Bennett (Jul 19 2010, at 17:26)
It baffles me that the Google market has such an abysmal web site. I want to find and review things on my computer rather than the tiny screen of my phone, yet I can't.
[link]
From: Paul Hoffman (Jul 19 2010, at 18:11)
Having a good community of raters doesn't help either. The other day I was looking for an iPod Touch app for Conway's Game of Life. *All* of them were rated 4+, even the one that only allowed a 20x20 grid, and the ones that didn't let you save or import configurations. The usefulness of those reviews was *way* below that of Amazon for music or books.
Worse yet, the links to the manufacturer's web sites were often useless as well. A number led to the home page of the ubiquitous "we'll do your next iPhone app for you" folks. Yes, I was only going to give them 99 cents, but the fact that they couldn't even be bothered to document the features of their app shows that you can't even mine the vendors' sites for useful information.
[link]
From: Jean Jordaan (Jul 19 2010, at 19:04)
Use the StackExchange platform. If it can sort the gold from the dross for millions of opinions, surely it can work for apps too.
[link]
From: Ronald Pottol (Jul 19 2010, at 19:59)
Well, I use http://www.appbrain.com/ it lets you pick and organize apps in your desktop web browser, then you fire up their android app, and it opens your selections in the market, so you can install them. Better than using your phone for the whole thing anyways.
[link]
From: Ciaran (Jul 20 2010, at 01:34)
Even the most basic filtering would go a long way to making things more findable in the Android Market.
I would particularly like to be able to limit searches to a particular license (e.g. GPL), or exclude things that contain advertising. I suppose neither of these things are going to appeal to Google though.
Google sees apps as either Paid or Free. I see Secret-Source, or The-Other-Kind-Of-Free. There is no correlation between these two sets of categories. Paid, with source available, is fine by me. Free, but not free, less so. But it depends on what kind of app it is.
More detail when browsing the list of results would be good too. Say I search for a battery widget - I definitely do not want any of the ones that require 'Full Internet Access' or other permissions I know are unnecessary for the application, but the only way I get to find out which is by trying to install it.
On a separate note, user comments on applications can be very useful, but they suffer from YouTube syndrome - i.e. 99% of those making the comments are complete and utter idiots. This idiocy is incorporated into the star ratings too of course. (An example that sticks in my mind is a comment, and 1* rating, on the excellent SwiFTP, which read "wut is this it dosnt do nething")
The comments seem to be simply displayed in chronological order. They need to be ratable in themselves, with the 'best' comments floating to the top.
[link]
From: KT (Jul 20 2010, at 01:47)
Google or Apple or Nokia should definitely come up with better app store search interfaces, otherwise it is hard for end users to find quality stuff. This guy tried to solve the problem, but Apple's red tape didn't allow it : http://bit.ly/alpJXe
[link]
From: David Shellabarger (Jul 20 2010, at 09:01)
Please go talk to the Market people next door.
I know this is a hard problem, but the current state of the Android Market is horrible.
Synonyms don't work.
If you search for "SmartLock" you won't find "Smart Lock". (you will now because I've added that keyword)
The app description is limited to 325 characters. Search is much less useful if you can only index 325 characters per app.
I know Google is going to vastly improve the Market website in the future, but there are some very basic things that could help the market right now.
[link]
From: Haik (Jul 20 2010, at 11:16)
Try typing this into www.google.com.
site:itunes.apple.com vancouver travel
iTunes app pages are indexed by google and there are methods for taking a user from their web browser to the app in the App Store on the iPhone or to the app page in iTunes on a PC or a Mac. A click away from buying or downloading it.
These app pages are linked to on the web. App reviews often link to the iTunes app page directly. So google's ranking and search magic applies to them as well. Apple solved the problem using the web.
The Android Market should copy/adopt the same method. i.e., each Android App having a single web page on the Android Market.
[link]
From: JulesLt (Jul 20 2010, at 17:14)
I'm not sure that Amazon does do a great job - it certainly does a worse job than a classic specialist retailer, and generally speaking I think it has the same flaws as the App Store - it tells you what is popular, not what is good.
The 'Other people who bought' can be interesting when you are in a reasonably specialised niche - but is pretty much useless at the top.
What it fails to do is, say, recommend music to someone who has bought a U2 or Madonna album that may be something they would like, musically.
A contrast would be something like emusic, which uses the semantic tagging data added by users to recommend similar artists (roots and followers).
Alternatively, this is why specialist retailers and magazines have evolved - music, literature, of software (even gaming software) - tends to split down into subcultures to manage the vast amounts of content - pick a genre, even one as obscure as Cowboy Zombie films, and you'll find a niche of people willing to construct a canon of the best and worst.
"Can something as useful as Amazon be built without a vast substrate of cultural commonality?"
My bet is no - there's a lot of anthropological and sociological research that suggests not. At the other end you have critics like Roger Scruton holding up Wagner as the absolute objective epitome of human artistry (the concept of an absolute canon of Art) - anything below that ideal means some kind of relativism (of judging art - including software - with regards to its audience).
And that is where peer recommendation engines fail - the person rating Trembling Bells or GeoDefence as 2 stars is evidently not my peer. A person who rates iFart and Lady Gaga at 5 stars is also not my peer.
What I guess I want is a service that provides recommendations from people who rate the same things to the same degree that I do (filtered by obscuring negative responses by people who rate things negatively that I don't) - but with a soupcon of randomness to ensure there is some element of new blood - although I find peer recommendation engines become an echo chamber (on Amazon I see the same items recommended) and I would rather be pointed to something good in a genre I do not know much about - say Burial or Dusk+Blackdown - that yet another psychedelic folk band - even if my tastes are more to the latter.
Historically, I think this is one reason why specialist retailers and magazines have evolved - understanding there is a difference between the preference of casual and hardcore gamers, between fans of indie-rock and 60s revivalism, etc - but in all these cases there tends to be an internal tension between conservative and progressive factions.
So far, the online markets tend to be poor at representing this. iTunes or Amazon are Tower Records - 'everything' is available, but as a retailer there is relatively little knowledge or active selling of the stock. A contrast would be something like Boomkat or Norman Records, where every release has a personal review by the staff (which is useless to the outsider, but useful once you understand the voice of the critic).
What I think needs to happen is for the distributors - the providers of the actual digital downloads - whether we are talking music or software or books or film - to become exactly that, and to allow new retailers to emerge on top of their distribution platforms - the equivalent of the specialist Crime and SF booksellers, or the games stores and magazines that cater for the Gamer audience - rather than trying to flatten all these audiences into a single channel.
I think there is a good reason why those patterns emerged (specialists vs general stores) - and recall that general stores generally predated specialists.
[link]
From: Peter Meyers (Jul 21 2010, at 11:44)
Not a solution here, but a data point. One of the last books I commissioned while working at O'Reilly was a title that many people scoffed at: "Best iPhone Apps" (http://amzn.com/059680427X) -- basically a catalog where the author profiled 200 or so fun/useful/etc. apps. It sold quite well. People need help with this stuff.
[link]
From: AceOfThumbs (Jul 21 2010, at 14:16)
A simple solution would be to collect anonymous information about usage of all apps and rate them according to the amount of time each app is used, then allow filtering by categories and keywords. People will spend the most time with good apps.
Boom, problem solved.
[link]
From: Chris DeVore (Jul 22 2010, at 08:23)
Great analysis, Tim - I agree the Amazon approach (others who bought that also bought these) is a productive direction to explore, as is the "discovered content and links approach" that AppsFire is using, and that was actually pioneered in this space by AppStoreHQ (ref. http://blog.appstorehq.com/post/329508024/apprank-whos-the-fairest-iphone-app-of-all )
AppStoreHQ has been working on Android app discovery for a while now (a much harder problem than iPhone thanks to the lack of canonical URLs) and we have an alpha of something I'd love to show you that combines both of these approaches in a new and effective way. If that's of interest please email me - chrisd at appstorehq dot com - your feedback would be immensely helpful.
[link]
From: BoD (Jul 22 2010, at 09:05)
Please also see this petition addressed to the Market team, for a list of things that *need* to be fixed on the Market:
http://www.petitiononline.com/androidm/
[link]
From: wendy richards (Jul 22 2010, at 13:07)
mimvi is the perfect site to address this mess:
www.mimvi.com
[link]
From: JulesLt (Jul 25 2010, at 01:26)
Few more thoughts - I think one problem recommendation engines haven't figured out (yet) is that quite often we work in cycles.
i.e. a new musical sound, type of game, or twist on a familiar genre emerges.
Eventually we exhaust our own appetite, and on the other side there is typically a proliferation of second and third rate copies - consider the number of first-person shooters on the PC, or Tower Defence games in the App Store.
In culture, about the worst act of timing you can have is to produce an absolute classic in a genre people are becoming board with, but not late enough for it to be the start of a revival or new wave.
Later on, of course, these items tend to get promoted as 'lost classics'.
(The same cycles also seem to exist in programming culture - timing is important).
Where does this fit in with recommendation engines? I guess my point is that they don't tend to reflect these cycles very well - they tend to reinforce whatever groove you are locked into.
Sometime I'd like to able to tell Amazon that it's not that I don't like Alistair Reynolds or space opera generally, it's just that I've had enough for now. Or that I'm rating a book as 3 stars because I don't like a particular book or series by an author I do enjoy.
[link]
From: Carl Jameson (Jul 30 2010, at 07:10)
One of the things the Apple Store did right was allow you to buy apps from anywhere in the world. I cannot understand why this is limited.
I have been buying stuff internationally for a decade without any problems (apart from shipping restrictions). I want to pay for quality apps! I suspect Google doesn't fix it because they want to establish Android as a platform where apps are generally ad-supported, strengthening Google's advertising power. However, watching the launch of Nexus suggests it's more likely that Google is incompetent and not cunning when it comes to international launches.
Regarding selling apps, I guess tax and legal issues may cause problems, but Apple managed to solve it quickly. Also, it's a solvable problem for developers since it's fairly hassle-free to set up a company in the UK even though you're not based there.
Limiting buying is a show-stopper though, and forces me to be an Apple app developer for earning money on apps since I think ads detract from the all-important user experience.
[link]