Hi guys. Thanks for the feedback and discussion. I'll try to reply to everything. Sorry for any messups in the following.
[snip - Matthew Toseland] Available in: If I just include a list of languages with largish l10n files, wouldn't that be original research? I haven't done anything with this yet. Tor doesn't have such a header. [/snip] Shouldn't we just list the languages we fully support? (At the moment English and perhaps a couple of other ones?) [snip - Matthew Toseland] Statistics: I agree the old stats are ridiculously out of date, but what to replace them with exactly? As far as I know there aren't any content stats; the FMS board stats for example don't break things down by content. Please post the stats you would like us to cite. They would need to be mirrored onto the web so we can have a somewhat permanent link on wikipedia to them. [/snip] No matter if we have replacement stats or not, I'd say wipe the old stats. They are misleading in my opinion. The FMS stats (USK at wOJh0RCFHn7o4tGNCJsvabwaOxEsW43XbP2sg~CUTEs,hSb98zf-5gOC18Ed4nzTtmvhkktK08phJav63sh04o4,AQACAAE/fmsstats/-1/) show the historical development in the amount of FMS communication. That is one possible source. Perhaps the author can help with more long-term and/or detailed versions of the stats/graphs. I don't know if the stuff at "Freenet Network Maps" (USK at Fggh8kKs-SLXOv23RILgsb1ITdV4IsLaRINkoRqvuqs,ysBW1CZ93Dv~jZB2tKa0kKW22VDQuw-YQkPx-D53L1U,AQACAAE/freenetmaps/5/) still works (on opennet perhaps?), but comparing graphs over time could perhaps be used to illustrate the connection strenght of the network. I remember seeing some content stats somewhere as well, but I don't have a link atm. One of the spiders could probably be tweaked to count file extensions and size as well. Shouldn't be a big task. [snip - Matthew Toseland] Google: We have a pretty good ranking in any case. There's an entire industry dedicated to optimising search engine rankings, we can't really afford to play with that, and IIRC Google ignores tags nowadays anyway? [/snip] I agree. I wouldn't spend money on that either. I didn't know about Google ignoring tags completely nowadays. [snip - Matthew Toseland] Website front page: [...] Adding graphs could be counterproductive, we don't want it to look like the Statistics page! We could mention the last commit or commits in the last week in a box somewhere maybe... The fundamental problem is that actual Freenet users get their news from the mailing lists or from FMS, *not* from the web site. This isn't going to change. We could post new builds on the website using the same text as goes out in the announcement... would that help? Usually it's just a bunch of bugfixes etc, of little value to anyone not on Freenet. And the 0.7 announcement would have to disappear. [/snip] I don't care too much about which of the above you would add, but rather that *something* is added to bring life to the website. I know that Freenet users get their news from Freenet itself - I'm concerned about the not-yet-Freenet-users. They won't ever get to the point of getting news from Freenet if they don't actually become Freenet users. And for them to become Freenet users, I think a "sneak peek" of some kind would be very useful. If you don't "sell Freenet" to the random users surfing by, you will probably rely completely on the existing users to bring new users into Freenet. [snip - Matthew Toseland] Alternatively, we could try to reinstate regular status updates - once a week, once in two weeks, or once a month. These could be posted to the web page as well as the mailing lists/FMS. This is probably the best way forward as it has a wider value. [/snip] That would be a good idea in my opinion. And find as many regular sources of news as possible. Something like a mix of the World Wine News (Example: http://winehq.org/?issue=352) and the Ubuntu Weekly Newsletter (Example: https://lists.ubuntu.com/archives/ubuntu-news/2008-September/000170.html) perhaps? [snip - Matthew Toseland] What is wrong with the prominent "Press enquiries should be directed to Ian Clarke", exactly? "Note to editors" etc are common on both web pages and press releases, no? We *could* move it to the People page, but we'd likely have more misdirected enquiries/frustrated journalists, no? [/snip] The users won't care about it. Journalists will surely be used to looking at "Contact"/"People" pages. They are paid to get as much information as accurate and as quickly as possible, so I'm sure they will bother clicking on the e-mail link marked with "Press enquiries should be directed to Ian Clarke.". The note is fine, just move it to the footer or the People page. I can't see why every single normal user have to spend time reading that line when not a single of them will care. [snip - Matthew Toseland] How much knowledge of p2p is it safe to assume on the part of a user? Most people visiting the web site will probably have come across the term "node" somewhere - e.g. supernodes in p2p networks. It's not immediately obvious how to cleanly juggle the second and third paragraphs to define "node" before using it: It would of course be convenient to define "node" in para 3, but we don't really want to have to move para 2 to after para 3, because imho para 2 is more important, isn't it? [/snip] I honestly don't think you can assume that the user knows what a "node" is. Most users of traditional p2p applications most likely won't know what a "supernode" is either. I don't know if it is a common English word, but as a foreigner, I had no idea whatsoever what a "node" is before looking I looked it up. Just add some kind of explanation of the term before p2, ala "By installing Freenet, your computer becomes a Freenet node, taking part of the responsibility of routing traffic between all the other Freenet nodes in the network.". [snip - Matthew Toseland] Why do people reading the What is Freenet? page need to understand "advanced stuff" ? The first paragraph is IMHO essential, it's not realistic to expect people to read the Philosophy page before reading the What is Freenet? page: Freenet *is* censorship resistant, that's one of its defining values. Having said that, it might be a good thing to include the diagram - if we can prune enough text for it not to go over a page. But we have to get across the idea of a censorship resistant distributed datastore, and the current text does that. [/snip] What I'm trying to say is that the page is very... Abstract. Not that the information is useless in any way - I completely agree that the information is important - but I believe it is very important to explain the practical basics of Freenet to new users as well. The goal of Freenet is very important, but so is the actual "product" and how it works. [snip - Matthew Toseland] What is the problem with the link to faq.html#firewall? The link works, and leads to useful text, no? If we move the ports to the Connectivity page we will need to update the FAQ item. What is the problem with the links for additional info on each OS? That is, what would look better? [/snip] The main problem is inconsistency. One link expands additional information while another links to a different page. The FAQ entry of "If you have a firewall or a NAT box click here for some info." could easily be added using the same expand effect as the installation instruction links. The problem with the installation instruction links is that it isn't obvious that the links expand additional information, and once the information is expanded, it isn't immidiately clear what information that have been expanded and what information that are part of the original page. A possible solution would be to align the 3 instruction links horizontally on their own line and somehow make them look like a "tab interface", making it clear to the user that he can switch between the 3 sets of instructions. Obviously this would require a bit of scripting and styling to do. [snip - Matthew Toseland] I have rewritten the post-install section, comments are welcome. [/snip] Much better in my opinion. Make sure to point out that the /wizard/ URL is only supposed to be used once though. [snip - Matthew Toseland] The security warning from Java isn't ideal, agreed. Fixing it would mean buying a code signing cert for a large amount of money per annum. [/snip] The main problem is that it fires off the keyword "insecure" to the user - which isn't too ideal for a project delivering free "secure" speech. It appears because of the .jnlp launcher, right? If so, perhaps the old-style downloading of an installation file is a better solution. A matter of taste I guess. [snip - Matthew Toseland] The rabbit: You mean in JWS or in the installer? We have a scalable version of it somewhere... [/snip] The very first dialog of the installer (language selection). Its not bad - its just not as good as it could be ;). Might as well present the user with a proper looking logo the first time he sees it after clicking "Install". [snip - Matthew Toseland] Improving the danish translation is of course a good idea, well volunteered. :) Seriously, if it's readable, we should keep it, right? [/snip] :P - Yes, we should. If I get the time, I will see what I can do about it. [snip - Matthew Toseland] I have made the auto-start option top-level so it is between "base" (which despite being mandatory had and maybe still has some optional components), and "plugins". If we want to keep the selection then we should show the accurate size of the plugins. [/snip] [snip - Matthew Toseland] Official plugins are loaded from the web site over HTTP, I don't think they are updated by default but I'm not sure; this is the main reason why somebody wouldn't want the plugins. Maybe saces can get around to fixing this by implementing revocable plugins over Freenet? [/snip] Yet I don't recall the installer informing me about downloading them off HTTP? Thats kind of a security problem, huh? Or perhaps they are included in the "offline" installers? [snip - Matthew Toseland] Apart from that, several of the plugins do things that advanced/paranoid users may object to: if you are on an untrusted LAN, MDNSDiscovery and UPnP are a bad idea; JSTUN may be a problem if you're worried about people trying to find your node by analysing STUN requests. [/snip] I agree - but we are in the installer here. This stuff logically ought to be located in the fproxy wizard as I see it. The plugins should probably be "installed" anyway, and then be enabled in the wizard according to what the user wants enabled and what not. [snip - Matthew Toseland] The current autostart detail text: "Start the node automatically on startup. STRONGLY recommended. If this is turned off Freenet will be much slower: it will take a long time to get up to speed after every restart, downloads will take longer, and the node will be much less useful to the network." [/snip] I still think it is important to inform the user about the installation of a system service. Besides that, good explanation. [snip - Matthew Toseland] Making a full-page question out of it would probably involve making significant source level changes to IzPack, so isn't likely to happen. We could get rid of the other options but is that a good thing given we need the page anyway? [/snip] I guess IzPack isn't as flexible as NSIS? Unless there are other good cross-platform installation platforms, then I guess we'll just have to do with the current method then. [snip - Matthew Toseland] I really don't understand why creating a user - a measure which is solely intended to improve security by not having Freenet run as System - causes concern for so many windows-using geeks. You can't log in as that user, and thus it isn't visible on the login screen. And it's standard practice on unix. And it improves your security against a hypothetical exploit in Freenet. WHY IS THIS BAD? Having said that, it does seem to cause some installation failures, so maybe we should turn it off. However, running as the installing user isn't straightforward, we'd probably end up running as System. Nextgens??? [/snip] I know that the user culture/standards are different for Unix/Linux, but on Windows, running processes as another user than the current is considered a mess most of the time. Windows doesn't exactly handle this stuff brilliantly. As others have pointed out as well, running running Freenet through the start menu "Start" group seems like a better idea. That would also get rid of all the problems with running Freenet as a system service, and gives the user easy control over the startup of Freenet. Geeks can move it to a system service if they want. [snip - Ian Clarke] How many Windows machines really have multiple active users anyway? My guess is a very small minority, we aren't talking about 1970s-style time-sharing mainframes here! Is there really likely to be a significant impact here if we just make this run for a single user? I agree that the security implications of allowing any user of a laptop to access FProxy probably outweigh the potential loss to the network of Freenet nodes when someone other than the person that installed Freenet is using the computer. [/snip] ... and ... [snip - Michael Rogers] Well, we can! provided we don't use the windows services at all but a shortcut in the startup menu or something like that... but we don't want to because that's user-specific. We want to maximize the uptime of nodes, not to restrict it to the timespan a specific user is logged on the system. [/snip] I agree. Most Windows machines (especially laptops) will be personal. Some might have extra accounts for family members and friends etc., but the main usage will happen from a single account. At least in the rich countries. By starting Freenet through the start menu of the installing user we also fix several of these problems. First of all, the node can either refuse connections from users different from the user running the node or place identify information inside the users profile folder allowing sharing of the same Freenet installation between multiple users. Furthermore, the node will start/shutdown with the user, thereby not affecting performance for other users on the system (and really: No matter how much we want node uptime, we shouldn't try to get it by running the node at times/places we aren't supposed to). [snip - Michael Rogers] Uptime is really important, I'm not disputing that, but we shouldn't try to get it in ways that are likely to piss people off. [/snip] I completely agree. [snip - NextGen$] Huh? What are we talking about here? If someone you don't trust has physical access to your computer you are doomed in any case... whether freenet is running or not when he gets his hands on the keyboard doesn't change anything. [/snip] There is another point: I might be trusting my family/friends/etc. in whatever stuff they do on Freenet, but that doesn't mean that I'd like to share my bookmarks, history, links etc. with them. In a multi-user environment Freenet shouldn't keep things shared (unless specifically asked to). [snip - Matthew Toseland] The custom firefox profile is essential. [...] Or maybe FF1? So it may be possible to remove the "don't close me" window. [/snip] I'm not saying we should remove it (I think it is a good idea - at least until we have the "private browsing" stuff in the next FF), I'm just saying that we *should* inform the user about it before installing it. It is the user's system - so we shouldn't mess around with it without asking for permission. I haven't personally spotted the "lockdown" bug that the "don't close me" popup is supposed to avoid, so I don't know. It's worth figuring out though, as the popup is quite annoying to users with updated systems. [snip - Matthew Toseland] Resurrecting the tray icon is a good idea, but hardcore gamers are going to want to shut that down as well, especially as it would have to be coded in Java for it to be maintainable. A minimal GUI java app only takes 10MB of RAM, so hopefully the majority would use the tray icon for starting and stopping Freenet. But even here, the majority of XP desktops have too many tray icons to show them all at once, so most users won't see the tray icon unless they go looking for it. Popup notifications make sense; exposing notifications for e.g. RSS feeds has been in the bug tracker for a long time, maybe exposing them skype-like makes sense too. [/snip] Let them shut it down then. If the tray icon has shortcuts ala: "Browse Freenet" ---- "Stop Freenet node" / "Start Freenet node" (Automatically switched depending on current state) --- Exit Freenet tray ... then none of our current shortcuts will be needed. Just add "Freenet tray" to the start menu. I'm not sure how Windows determines which tray icons to hide (when auto-hide is enabled, as it is by default if I remember correctly), but new icons most likely won't be at once. Adding baloon-notifications for updated freesites/etc. should probably un-hide the icon as well. If the user uses the icon regularly, it will probably be kept visible by Windows. [snip - Matthew Toseland] A free public bandwidth testing service might be an option but would probably be detectable, likely easier to identify as Freenet than JSTUN is?? Something hosted by us would certainly be detectable as Freenet-related. [/snip] If added with a text ala "Freenet can connect to a public bandwidth-metering service to determine the speed of your connection. Close down all other bandwidth-intensive programs (Downloads, file sharing, ...) and click "Auto-detect" to do so", it should be fine? I don't think we should make our own, but rather find a suitable public service (if possible). [snip - Matthew Toseland] - Java doesn't provide a way to detect the total AFAIK. It might be possible to build platform specific glue to find out. [/snip] Odd. Ideally, Freenet would auto-adjust RAM according to the amount free at any time. So if the user launches a RAM-requiring game, Freenet would detect the lack of free ram and free some of its own. When free RAM is available once again, Freenet would allocate up to a certain percentage again. [snip - Matthew Toseland] Ian wants to get rid of activelinks. Any thoughts? IMHO it should be decided by the theme... the theme we are considering for the new default, by Dieppe, will not have them. http://amphibian.dyndns.org/freenet/browse-mockup/html/browse.html http://amphibian.dyndns.org/freenet/mockup2/mockup2/html/browse.html (Probably somewhere between the two; both by Dieppe, I just host them). IMHO activelinks are an important part of the first time user experience: they reduce the chances of the user not being able to see what to do next with Freenet. On the other hand, Ian thinks they're ugly. And they do make the page take longer to load. I would be interested in your views: we can reopen the debate on activelinks if need be. [/snip] IMHO, the whiteish colours we have atm are *much* nicer than the blackish ones. The general layout of the first mockup is way better though (yet has way too much spacing atm). I generally think activelinks rock. They are kind of unique to Freenet, they are pretty (mostly), practical and unified across Freenet and the community seems to like them. In a marketing view we ought to protect that unique-ness - not remove it. Perhaps they could be upgraded a bit. Perhaps by creating a bigger default size closer to standard banner size to allow both a title, logo and a short description of the freesites behind the logos. That would even allow authors to (auto-)generate activelinks with the latest freesite news delivered straight to the fproxy homepage. Bookmarked freesites ought to be cached by Freenet, thereby also eliminating the loading time problem with fproxy. [snip - Matthew Toseland] Widening the menu so that each item fits on a single line would likely waste a lot of space with the default theme, especially as in many languages these are wider still. Introducing some spacing might help, but the real solution is to eliminate unnecessary menu items, use submenus, and a new theme, probably with the main entries in a menu bar at the top. See above. [/snip] I agree. But you could easily make a quick fix ala: Browse Freenet Messages Plugins Connections -- Friends -- Strangers Transfers -- Downloads -- Uploads Statistics Connection [snip - Matthew Toseland] The messages page only shows up if there are messages, no? And when you remove the last message, it should disappear from the menu, and redirect to the Browse Freenet page?? I'm sure that's how it worked - I wonder what has changed??? Are you sure about this? [/snip] No, the "Messages" menu entry is always present. You are redirected to the main page after deleting the last message, but the menu entry remains (and shows an empty page upon visiting). So the interface is a bit cluttered by having an (mostly) empty page in the menu at all times. [snip - Matthew Toseland] [about the "Add node" parts of the friends page] With regards to the form, is it sufficient to simply mention it? [/snip] That could do. [snip - Matthew Toseland] The thinking with the ports box was that adding a Friend would often not work because of port forwarding issues. That was before we had an "Internet connection" page. Should we completely move it? Or maybe have some more self-documentation? "Please note that adding a Friend may not work unless you have forwarded UDP port XYZ on your router" ??? [/snip] Port stuff shouldn't be cluttering the "daily interface" in my opinion. Move it to the "Internet connection" page would indeed be a solution. More self-documentation is also useful. Actually, the user should only be "recommended" to forward ports if they are not (we know if we are forwarded, don't we?). [snip - Matthew Toseland] Maybe we should get rid of ALL of the dev flogs. I know when I have updated mine a number of people have been offended at its being included. And I very rarely update it - after all, I have better things to do with my time, like developing Freenet! This is evidently true of the others too. [/snip] The inactive ones at least (which, well, are most of them). I don't think including dev flogs are a bad to start with. It seems like a very good way to serve initial content to new users - but that of course requires the devs to actually post something on their flogs on a regular basis. Otherwise they *will* only serve to disappoint new users. [snip - Matthew Toseland] ISF has a description in English saying it's a french index. Isn't that enough? [/snip] I didn't note the hover-text when writing the review. Perhaps just add that hover text in a set of parentheses as with the Textlink/Activelink indexes? [snip - Matthew Toseland] [About indexes] [...] - French flogs will be of interest to some users. Porn will be of interest to some users. Isn't much of the point of the indexes to help the user to find something of interest reasonably soon after installing Freenet?[ ...] [/snip] My main concern is that the indexes blindly index every single freesite out there with equal weight. That makes it time-consuming to find the "quality" content. Some of the indexes sort away some of the most useless stuff, but there is still a lot of "trash" indexed. Its not that I think the indexes, for example, should exclude all porn, but indexing a single-paged Freesite with 10 generic internet-porn-pics without any text at all seems stupid. Surely people looking for that can plug in some dirty words in Google and enjoy their life in that way. What I miss is sorting out the "worst quality" content no matter the topic. Obviously that is up to the index owners to decide in the end. [snip - Matthew Toseland] - I agree that topic-specific indexes are the way to go long term. However the web at large seems to think that search engines are the way to go. Hopefully we will have both. Right now we probably don't have enough dedicated users to maintain more than one or two topic-specific indexes. [/snip] As far as I recall, indexes/portals came before search engines on the real internet. Probably for good reasons. So following a similar path for Freenet would seem like a good strategy. I'm hoping to get enough free time to launch a portal I've been working on, but I can't make any promises at the moment. - Zero3
