Just a collection of things that might be worth implementing. Maybe giving someone ideas :)
After reading this article on AI and how it helps propaganda, I wonder if one could make a counterproject analyzing news, and trying to understand trends and what the propaganda machine is trying to tell. The influenced people suck up the propaganda eventually, but probably without noticing it. How about algorithms seeing these trends and attempts for propaganda before we notice?
This post on languagelog shows the distribution of the word “this” in America. Would be fantastic to see such distributions over the world, with a freely selectable search term.
4k resolution tv are getting increasingly cheap and common, but HDMI2 output on computers are not. I can use HDMI or Displayport1.1 output on thinkpads, but only for 1900×1080 resolution. The monitor can play 4k video offered as mp4 container via DLNA. How about creating a 4k screen on the computer, encoding it like a video and offering it via DLNA to the tv?
Moving into a very small flat in Tokyo, with no good view outside? Ideas of using computer screens to replace windows are coming up. Most interesting is to collect content: waterfalls, nature and so on, free content. Kickstarter projects are also around: https://www.kickstarter.com/projects/atmoph/atmoph-window-your-room-can-be-anywhere
http://www.kanjidamage.com/ and related books look like the sane approach to learn Kanji. There should be a free version of kanjidamage.com which
lang-8.com is a great website, one can write texts in various languages and get them corrected. In return, one corrects others texts. A karma system helps to spot if people are contributing or only expecting others to check their texts. Ideas which could help to improve future implementations of lang-8:
Imagine having media files like pdf, mp3 and videos on multipe computers. You watch the first half of a movie on the desktop, later you enter the train. When starting to play the same video file, it should continue at the place where I stopped before. Same for audio files and pdf. This should be done in syncing the current “media place state” to a server process which always runs. The media players like mplayer and evince would need how to deal with this, for storing the current “play state” and also fetching the state to continue with playing. Apple started to provice this functionality, but there is nothing opensource as of now, AFAIK.
This xkcd++ would be worth implementing: http://xkcd.com/350/ .
How much of a blockdevice has to be changed so the filesystem can not repair it? If a single bit gets changed, how are chances that this one bit will lead to lost data? Or even go unnoticed? One could in a blockdevice modify single bits, and then verify a) if mount notices something b) if fsck notices something c) if files ontop change silently. Comparing this for multiple filesystems.. is interesting.
Might be interesting to take a unicode font, and then have a look which chars from unicode it implements. Then do a graphics demo like BB (debian: apt-get install bb) with these chars. They should provide many more possibilities to match the actual picture.
With prism users would like to have the broadcast capabilities of facebook (conveniently post to a group of people), while having the option to stay anonymous for others than the recipients, and to encrypt the broadcast so only the recipients can read it. Writing an email to recipients and encryping with the recipients pubkeys works, but is not as convinient. diaspora is an approach into a decentralized direction atleast, and doing this in an open way. Maybe just a graphical frontend for mail+gpg is a good start.
Everybody has books around from childhood: books explaining things in simple words, and simple storys. Those books are not of big value to us, they are in danger of beeing thrown away without fulfilling a further purpose.. but this kind of books is important to others learning this language! Hm.. one could setup a web portal to exchange such books. Doing the environment a favor, and helping each other to learn languages!
Nested KVM allows to run virtual machines in other virtual machines. Testing things like OpenStack gets easier: create an image, and inside several virtual machines with the OpenStack components.
puppet implements manifests on client systems. This workstyle means a paradigm change for sysadmins. How about this:
http://virtualthreads.blogspot.com/2006/02/understanding-memory-usage-on-linux.html describes how to see what memory/libs are used/among what processes. This could be presented nicely graphically, imagine a bubble 'libz.so' and seeing what processes use it..
There are kickstart-files for redhat* distros handing out informations how to setup a linuxbox. Suse does a similiar thing with autoyast, the file using a completely different syntax thou, and admin unfriendly xml format. Ubuntu uses preseed-file, containing simply the selections that are made in the installer, debian/fai uses a more complex format with a directory-structure mounted and avaluated. There could be one ontop-standard describing a partitioning, what users to be created etc. and optional parts ontop that get only nonoured at the distro. Such a default-setupdescription could also serve for bsd, solaris deployments.
This network hosts data of people and interconnects it, you have to pay to see who is looking you up. Such funktionality could be provided in an open way. People should host their profiles on a website and interconnect to other people in linking to their profiles - located on other URLs on the internet. Lookup semantic web. A network would span over the URLs. And my data would also be hosted by me, thats what i want. Ofcourse people would also be able to host their data without own servers, there would be providers for this, also free ones, as for openid now. The idea settled reading a thead at the tlug.jp mailinglist archived here: http://www.tlug.jp/ML/0809/msg00082.html .
Update: there are now at least APIs like http://code.google.com/apis/opensocial/ to communicate such informations across borders of webapplications like xing and linkedin.
ipxe/gpxe http://etherboot.org now allows more flexible pxe-booting, i.e. one can fetch linux-kernels via http. There are services offering distros ready to get booted. Next step would be to directly boot a iso9660-cd-image around on some ftp-server on the intranet! The single parts are around; ftp-files can partly be read, also be mounted as filesystem. iso9660 can be accessed there blockwise for getting first blocks for the boot, thats just required in gpxe. With this new live-cds could be tried out onthefly over the internet.
Slack ( http://www.sundell.net/~alan/projects/slack/ ) is a small and nice distribution-system for configs and software, sysadmin-style. The author states in slack-intro:
I haven't invented the parallel universe simulator yet to see how the world will react to the actions of your scripts
Actually.. implementing this could be fun. Either making a copy of the whole system in a subtree with rsync and then running scripts chrooted on that (escaping daemon-restarts etc, as those would influence the real system) or virtualizing the whole system and running the script on that. After that diff'ing i.e. all files and report what was changed.
There is Second Life, kind of muds, kind of how people use world of warcraft. Maybe its a different more intuitive way to present some media and bring people together. But some points could be improved:
Recently visiting an old fellow at a research centre at Garching i saw they use video-chat on a daily basis. Also for things where i would simply prefer instant messaging. There are many things, all provided by one single app:
Hamachi provides an instant messenger, users can establish vpns connecting each other. Thats a great idea, just that hamachi isnt free but proprietary. Why has noone yet suggested a jap to implement this into jabber? Maybe using openvpn as backend?
This website works not bad for me, I can edit it online and the site gets available for me and others via search engines. downsites: It should also be available for offline usage thou. Also export to pdf should work.
Idea: modify the data locally as markdown file, have pandoc ( http://johnmacfarlane.net/pandoc/ ) create dokuwiki format, push into a git and to the webserver. Generate pdf if required.
No such package as of now.