March 19, 2013 at 7:54 pm #6412
Just thought I’d throw in a feature request:
Top 50 most downloaded apps
Weekly featured app (Randomly generated from those in the Top 100 that are not in the Top 50)
BenMarch 21, 2013 at 5:15 pm #6471
I think a new category “most downloaded” is even more useful and even easier to implement. In general I appreciate your idea.March 22, 2013 at 11:12 pm #6495
Yes, we need to make stats accessible in on-device client and use in all this ways and more. I’m sure we’ll get to that.April 12, 2013 at 4:09 am #6819
I’ve just created a feature request here: https://f-droid.org/repository/issues/?do=view_issue&issue=222. Would be interested to get some feedback on what an appropriate UI would/should be, because I’d love this feature enough that I will code it up if you can agree on an interface.April 15, 2013 at 10:49 pm #6857
Generally, I think that we should start putting up spec documents for app design/behavior in the wiki, to try to get whole (or it least sizable) picture first, because applying many small random changes won’t lead to good design/look/feel.
That said, I’d find following approach to consistent with what we have now:
We have “advanced features” setting in the client, which activates for example showing of Android package ID on app details page. And app details page seems just like right place for this and similar kind of info, and at least initial imho it should be enabled as “advanced feature”. For multiple reasons – because it may overload and skew unsuspecting users’ choices – for example, older app would have larger total download number, but that doesn’t mean it’s better. We can get round that by showing stats for last week or month, but then I raise my hand as advanced user that I’d have to see complete stats too. So, again, would make sense to show more or less detailed stats as advanced feature, and then see what sense to enable as “standard” feature either.
Besides d/l stats, I’d also like to see on app details page reasons for an app being marked as “incompatible” – actually, that would be higher priority for me. (But then again it’s blocked by https://f-droid.org/repository/issues/?do=view_issue&issue=227)October 10, 2013 at 11:53 pm #9007
While I was responding to a similar thread, I thought I’d bring the discussion here instead.
As discussed here or elsewhere, some of the issues revolve around privacy and also the accuracy of the download data.
However, another aspect is whether “most downloaded” is actually a good measure of “good” apps. The closest analogy I have experience with is the Wii Homebrew Browser – a piece of software that works like an app store for the Wii Homebrew Channel. They keep download stats, and rating stats. I created a chart here to see if there was any sort of relationship:
It is a bit of a mess. The only clear relationship is if you are in the top very small percentage of most downloaded, then you tend to have a good rating. Other than that, it is not very definitive. Or is it? I dunno.October 28, 2013 at 9:49 am #9447
My takeaway from that graph is that the number of downloads is not really a good measure of anything, but it could be that I’ve just interpreted it so as to back up my pre-existing beliefs.
My biggest concern about the “most downloaded” way of doing things is that it tends to be self-reinforcing – i.e. the thing that is most downloaded just becomes more and more downloaded, while the potentially better things languish. I think the graph can be interpreted in that way as well.October 31, 2013 at 2:33 pm #9517
I agree with @ciarang here – most downloaded may or may not be a good measurement, but even if it were exact we would have tons of apps that would be overlooked by our users.
I would personally prefer ratings. I’ve been coming up with an algorithm to rank apps, suggestions gladly accepted:
* Users can rank an app, said rank applies to the vercode installed
* Rank is posted to the repo, and we use some mechanism to check that the user didn’t vote yet (not sure how to do that)
* To calculate the overall rating of an app, the latest few releases will count (e.g. 0.2*apks[-3] + 0.4*apks[-2] +0.4*apks[-1] or something like that)
The only problem I see is how to come up with a way to make sure that people can’t vote multiple times on a single app-apk.October 31, 2013 at 2:34 pm #9518
Forgot to say… Apps would require a number of votes to make it to the rankings, or else a few 5-star ratings would kill it 🙂October 31, 2013 at 2:50 pm #9519
While I agree with @ciarang and @daniel Martí, that we should favor a rating based list, this algorithm seems to be particularly bad. If I give an app 5 stars I don’t want to vote again and again for every version. If the app becomes bad or better, I’ll change my rating. I think it is valueless if my votes fade away earlier or later. Thus I don’t need to vote in the first place, especially not when we can give comments.
Permanent votes and statements are great or it seems totally pointless. The problem with double voting can be solved with registration, except you have a technically more sophisticated solution.October 31, 2013 at 4:02 pm #9521
My biggest concern about the “most downloaded” way of doing things is that it tends to be self-reinforcing – i.e. the thing that is most downloaded just becomes more and more downloaded, while the potentially better things languish.
This is indeed valid concern. However, if such thing happens, you can go ahead and promote better alternative to fix situation. For example, you can write blog posts about new application (like SourceForge staff does) or even remove/archive old software in favor of newer and better supported.October 31, 2013 at 4:06 pm #9522
@AlexanderR: I don’t think that us making balance manually is a good idea. At an average of 5-10 new apps per week, we’d spend more time worrying about it and writing blog posts than doing actual work (not to mention that some developers would get mad about said ratings).October 31, 2013 at 4:10 pm #9523
@AlexanderR: Free Software implies the freedom of choice. We are happy to have several alternatives in some cases (like PDF viewer). It is the choice of every user to use to one, which fits most. If F-Droid ever starts censorship by removing alternatives, it would be time to fight F-Droid by forking. And I hope that would happen. So – we can be happy that this community is not driven by such bad ideas.October 31, 2013 at 4:30 pm #9524
not to mention that some developers would get mad about said ratings
That’s a good point! That’s why I think there should be a community review and registration, like in case of Mozilla. I’m the author of two Firefox/Thunderbird Add-ons. I remember once there was a really strange guy who gave me a 1 star rating and the comment that my Add-on wouldn’t work with his E-Mail client. Well the point was, he didn’t use Thunderbird, but a proprietary crap app and tried to force me (and by the way other developers too) to port it to that proprietary crap this way. Of course I asked for a community review (and the voting was deleted). Furthermore there should be a way to reply to comments and a notification for the developers, if they get a review. We should implement a way to contact the developers directly beforehand, since otherwise the comments will be spammed with questions.December 25, 2013 at 3:46 pm #9957
The problem with user opinions is that what shows up in user ratings may not actually reflect reality. For instance, if I develop a new app, all I would have to do is create 10 accounts, post ten 5 star reviews of my app, and all of a sudden I am at the top of F-Droid.
On the other hand downloads are based on reality. These can be rigged for a short while, but in the long term reality takes over. Some statistics I would like to see:
Top 50 new apps
Top 50 updated apps
Top 50 most downloaded apps (ever)
Top 50 most downloaded apps (last month)
Top 50 downloaded increase (last month)
Top 50 downloaded increase by percentage (last month)
These statistics would be valuable to different users. New users would want to see the Top 50 most downloaded apps of all times. Returning users would want to see new apps and the apps which are trending, and the remaining statistics show these results. All of these can be compiled from F-Droid statistics located here:
To calculate the Top 50 downloaded increase you would need statistics from a month ago which I do not have. Simply subtract last month’s downloads from this month’s downloads.
To calculate the Top 50 downloaded increase percentage simply take the number above and divide by the total number of downloads from last month.
Here is a list of the top 50 downloads of all time as of Christmas 2013:
1 org.fdroid.fdroid 3,620,715
2 org.adaway 1,025,356
3 com.fgrim.msnake 153,248
4 ru.o2genum.coregame 153,048
5 org.scummvm.scummvm 151,901
6 com.zagayevskiy.pacman 151,787
7 org.chrisbailey.todo 151,404
8 net.phunehehe.foocam 151,209
9 org.dolphinemu.dolphinemu 151,106
10 org.hanenoshino.onscripter 150,181
11 org.kde.kdeconnect_tp 119,109
12 org.mozilla.firefox 112,005
13 com.tum.yahtzee 111,892
14 org.ppsspp.ppsspp 110,856
15 net.i2p.android.router 71,618
16 org.torproject.android 66,818
17 com.fsck.k9 60,822
18 com.google.zxing.client.android 56,517
19 net.osmand.plus 55,081
20 com.owncloud.android 53,832
21 org.adblockplus.android 49,544
22 org.videolan.vlc 45,551
23 com.ghostsq.commander 40,154
24 org.jfedor.frozenbubble 37,909
25 com.morphoss.acal 33,675
26 jackpal.androidterm 32,447
27 dev.ukanth.ufirewall 30,357
28 com.gpl.rpg.AndorsTrail 28,695
29 cx.hell.android.pdfview 27,935
30 org.fastergps 27,486
31 com.elsdoerfer.android.autostarts 27,189
32 org.geometerplus.zlibrary.ui.android 27,063
33 com.andrew.apollo 25,783
34 org.thialfihar.android.apg 25,204
35 at.bitfire.davdroid 25,116
36 dentex.youtube.downloader 25,110
37 org.pocketworkstation.pckeyboard 23,695
38 org.coolreader 23,525
39 com.android.keepass 22,855
40 org.connectbot 22,639
41 org.openintents.filemanager 22,382
42 com.menny.android.anysoftkeyboard 22,169
43 com.eolwral.osmonitor 21,672
44 org.wikipedia 21,488
45 aarddict.android 21,325
46 info.lamatricexiste.network 20,064
47 info.guardianproject.browser 18,813
48 net.fred.feedex 18,169
49 se.johanhil.duckduckgo 17,832
50 com.googlecode.networklog 17,081
The forum ‘Development’ is closed to new topics and replies.