Modified: 12 March 2010, 11:39 UTC

sabik's blog

Entries marked: idea
(show: listfull)

10 January 2014, 12:30 UTCIdea: grape juice tasting

Idea: grape juice tasting for children. This one's a Riesling and this one is a Cabernet-Sauvignon blend and this one is a Sémillon.

(Not sure it's a good idea, or a wise idea, but it is an idea...)

PS: Apparently this already exists. So much for that one.

[permalink] ‣ keyword: idea

22 November 2013, 6:30 UTCSelf-driving model cars

I wonder if technology is now at the stage where one could put together a self-driving model car. One that uses a camera for lane-following, collision avoidance, obeying street signs and traffic lights...

A rough calculation suggests that a Raspberry Pi would just about fit inside a 1:32 model car. Certainly there are radio-controlled cars as small as that, and a computer these days is not going to be much larger than an older R/C receiver. It needs to be powerful enough to do video processing, but that, too, doesn't sound like a problem any more. If it's really a problem, one could use a bigger scale, maybe 1:18 or 1:12 or something; that would give ample room for pretty much anything.

I wonder if that already exists, or if I should try to put it together and popularise it as a hobby.

[permalink] ‣ keyword: idea

1 May 2013, 6:35 UTCGoogle Glass

Is Google Glass likely to be disruptive? Does it fall into the class of truly innovative products, or will it turn out to have niche use at best, like video phones and jet packs?

It's cool, to be sure; but is it cool in a world-changing way? Or is it, to borrow Bruce Sterling's mildly derogatory turn of phrase, a toy rocket-ship?

[permalink] ‣ keywords: future, idea

26 February 2013, 4:57 UTCDemocratic social network

The problem with social networks such as Facebook is ultimately that the incentives are all wrong. Fundamentally, they are answerable to shareholders, often at the expense of users. Users are the product.

Distributed, federated networks are one possibility, but they have their downsides. Many technical aspects are much more challenging than centralised (secure multi-party computation). They're more difficult for casual users (need to install SW rather than just visiting a website). In a centralised network, the centre can do things; lead, manage, steward; A/B testing, compelling, market-building things; represent the network as a single entity / legal person. It's a technical solution to a social problem.

Idea: Run a centralised network as a community project, with a foundation behind it and Debian-style governance. It would have the same structure as centralised networks, but with the CEO ultimately answerable to the body of the members rather than shareholders.

If successful, the president of the network would have mandate from ¾ billion people (current size of FB). Few in the world can say that... In some ways, a network like that really could take over the world, become the next generation of governance, lead to the foundation of a world government.

[permalink] ‣ keywords: idea, social

3 November 2012, 10:42 UTCExercise bikes and Google street view

There should be an app that wires up to your exercise bike and shows you Google street view...

Inspiration: Stationary bike weight-loss success story

[permalink] ‣ keyword: idea

29 April 2012, 11:01 UTCxkcd random²

If new ideas are new connections between existing ideas, as John Cleese says in that video that went around the other week, maybe one way to be creative is to randomly pick pairs of existing ideas until one combination clicks...

In that spirit, I present xkcd random².

Note: if normal Reload doesn't work, try Shift-Reload (or Shift-Ctrl-R).

[permalink] ‣ keyword: idea

19 April 2012, 5:13 UTCVirtualisation on the desktop?

Today installing a package is an all-or-nothing security proposition Ingo Molnar

Of course, we do have sandbox technology on Linux; it's called virtualisation. It's a bit heavy-weight, but sufficient as a proof of concept.

So, the basic functionality should be to put a PPA in its own sandbox.

I wonder whether that's a good idea or a terrible one. Would be interesting (but quite a bit of work) to find out...

Unsorted thoughts

[permalink]2 comments ‣ keywords: FOSS, idea

19 January 2012, 0:29 UTCGraphical programming languages on tablets?

I wonder if graphical programming languages might find a use case on tablets.

Inspired by seeing Turtle Art at the OLPC desk yesterday, with a touch screen on the OLPC prototype, followed shortly by someone else's tablet PC (actually an Asus Transformer). The juxtaposition hit me…

On desktop computers, it's always been ultimately more convenient to just type it in on a keyboard. Graphical programming languages have been tried, but the disadvantages have always ended up outweighing the advantages, with very few exceptions (such as the Turtle Art).

On tablets, though, typing is always at least a bit of a hassle, while dragging icons around the screen is the native mode of interaction. The balance might well turn out the other way.

I may have to see if I can put something together and see what happens :-)

[permalink] ‣ keywords: programming, idea

18 November 2011, 1:06 UTCIdea: self-driving toy car

I wonder if one could make a toy car that's self-driving, so that it goes around a play-carpet with roads obeying all the road rules... The technology is probably just about there, it'd be more than an Arduino project but well within current capability to do it all on-board (or you can use wi-fi and off-board processing).

[permalink] ‣ keyword: idea

13 July 2011, 2:24 UTCImperative-looking syntax for Mercury?

I wonder to what extent it would be useful to make a programming language that looks familiar to programmers used to imperative (possibly object-oriented) programming, but which is actually a logic programming language with a whole lot of syntactic sugar and syntactic salt (and maybe some syntactic cinnamon as well). With some care, it should be possible to make the obvious imperative-like things act in the obvious way, while not getting in the way of the logic-programming things where they're an advantage.

Of course, the real biggest problem of a new language is that it will lack libraries and frameworks. Given that Mercury compiles to JVM, perhaps the JVM libraries could be used... except they will be alien to the language.

Or maybe I'm re-inventing Erlang, poorly, although Erlang is based on functional programming rather than logic programming.

[permalink]1 comment ‣ keyword: idea

6 March 2009, 4:43 UTCIdea: to-do schedulers and GPS navigators

To-do schedulers, especially mobile ones, should integrate with GPS turn-by-turn navigation (or at least Google Maps) and take travel time into account when popping up reminders.

[permalink]1 comment ‣ keyword: idea

27 March 2008, 18:15 UTCElectrolytic Underwater Breathing Apparatus

Water can be split into hydrogen and oxygen by electrolysis. Normally, one would keep the two gasses separate; however, for the purposes of life support, they can be mixed and breathed!

Such a mixture would have ⅓ oxygen and ⅔ hydrogen — somewhat rich, but breathable for half a week at one atmosphere, or the total pressure might be reduced to 63 kPa for indefinite use. Underwater, it would be breathable on a short-term basis down to some 32-38m, just like air, though with completely different schedules and without the gradual impairment of performance characteristic of nitrogen narcosis.

The main advantage would be if a compact source of power can be used (for instance nuclear). The main disadvantage would be the risk of explosion. Indeed, the E in the hypothetical acronym might well stand for "exploding" rather than "electrolytic"…

[permalink] ‣ keyword: idea

4 October 2007, 6:05 UTC"You are here" on the Internet

I can't claim this idea for myself, it was a group effort at Conflux.

There should be a browser plug-in that shows your current location on the Map of Online Communities (or, failing that, the Map of the Internet, which has the advantage that it unambiguously covers the entire IPv4 space).

[permalink] ‣ keyword: idea

13 September 2007, 15:33 UTCReputation link, combat idea for distributed RPG

pfh pointed me to Andrew Clausen's pages, in particular his work on reputation and other ideas and projects. Haven't read through it all yet, but the titles look very interesting!

This is all in aid of my vague search for things that could come together to form a distributed MMORPG.

Thinking of it, reputation probably would be a good part of it. I wonder if you could use it for combat? I'm reputed to carry a +3 sword, you're reputed to wear dwarven mail. Roll for damage :-)

It'd be a strange world, where belief is very powerful, but that's not so wrong for fantasy. For that matter, if I can convince you I carry a +3 sword, I don't need a reputation. I can use it against you. Presumably that'd be one way to start a reputation. You'd probably need a custom of some sort of apprentice relationships, where an established player testifies to a novice's weapons, so that people can get started easily.

Even if it's not used for that in particular, it'll still be handy in various other places.

[permalink] ‣ keywords: link, idea

24 August 2007, 7:39 UTCPersonal currencies

I've been thinking about ripple monetary system and I'd come to the conclusion that it may well be even more revolutionary than Ryan himself presents it.

You see, he always talks about accounts in dollars. Now, dollars are defined country-wide and even globally: a dollar is a dollar regardless of who owes it to whom where. It's not quite true, there is such a thing as "doubtful debt", but for the most part the central idea of such global currency is to gloss over these complications as much as possible.

This has some advantages: in particular, it's possible to run such a system without computers, using physical tokens (coins and notes). But it also has some disadvantages: as LETS proponents will tell you, it's possible for local areas to have both supply and demand but too little of the country-wide currency to run the local economy efficiently, for instance. Yet even local LETS currencies are based on the same concepts as the global currencies, just in the small.

A system like ripple would allow personal currencies — currencies defined by a single individual, possibly even several for different purposes.

Perhaps they would be denominated "hours" or "favours", as in "Bob owes Alice a favour".

A transaction would involve a chain of exchanges along a path through the friend network between seller and buyer; each obligation would be local, but in total effect the network would likely be global.

Logistically, this would mean that all transactions have to go through the computer system, but that's not a big problem: on-line, it's no problem at all, and off-line you're carrying a mobile anyway, so it can do that as well.

I wonder how (and if) such a world would work...

[permalink] ‣ keywords: ripple, idea

15 June 2007, 13:37 UTCIdea: Megatokyo the ARG

Now *that* would be scary. "Meet half-way" sort of scary.

An Alternate Reality Game with a Megatokyo-like setting.

Fred Gallagher mentioned something at the natcon last weekend about how Megatokyo plays with perceptions, how different people see the same streetscape (Largo's vs Piro's, for instance; or Kimiko's — does she see the killbot, or just the fanboys?).

An ARG does the same, in a very different way — it deliberately creates a different perception for its players; though usually only really one alternate perception, and a limited one at that (especially limited in time — AFAIK no extant ARG is open-ended, they're all single-quest).

Combining the two, done well…

[permalink] ‣ keyword: idea

12 June 2007, 2:22 UTCTwitter and the Weasley Clock

Seems to me Twitter is only a step away, now, from making the Weasley Clock a reality...

All you need now is a clock with nine hands connected to nine servo motors, some programming and some bits and bobs and voilà!

(Of course, clocks and dials and gauges with servomotors would be interesting generally — you could probably make a good hobby/business selling those... Ages ago I read a story about someone who made a weather clock out of one of those, and no doubt if you could just buy the hardware, print out a clockface and go, a lot more people would use them, for all sorts of things...)

[permalink]1 comment ‣ keyword: idea

31 January 2007, 14:03 UTCZeroconf matchmaking?

With Zeroconf (Bonjour, Avahi) computers can advertise (and discover) various services on the local network, like a printer or a music folder. I wonder if there's software out there that uses this to implement social software, like the Lovegetty gadget or the toothing hoax?

Probably not so useful on a hard-wired network, of course, but with WiFi in public or semi-public spaces, why not?

The subtle distinction between advertising _toothing._tcp and _toothing._udp is left as an exercise for the reader :-)

[permalink] ‣ keyword: idea

29 October 2006, 4:20 UTCIdea: ingredients list

It might be interesting for a shopping site to let users convert their cart into an "ingredients list" that can be bookmarked or blogged.

That way when you're blogging a recipe or an electronics project, you could make the "ingredients list" or "components" section a link to the shopping site, which would offer a button to "add all these to the shopping cart".

I don't know if independent bloggers would use this (they might), but even if only the in-house recipe blogs use it, it would still be useful.

Implementation note: the ingredients list can be directly embedded in the URL, presented as an HTML snippet or you have to store it indefinitely. Embedding in the URL is the easiest but will result in excessively long URLs. HTML snippets are OK — people use them for random quizzes — but non-standard. Storing indefinitely imposes a burden; disk space is cheap, but it has to be on-line, ready to retrieve at a moment's notice.

[permalink] ‣ keyword: idea

21 October 2006, 8:07 UTCIdea: scrollbar temporary marker

It would be nice to be able to right-click on a scrollbar slider and set a "temporary marker" — something to keep your place while looking up something elsewhere on the same page. When dragging the slider, it would then function as a detent, or the marker itself could be clicked to jump back to the spot.

Perhaps they could be coloured, so that you could have a blue spot and a red spot, but that'd be a bonus. The basic functionality is what's important. Colours could be generated in HSV space by holding the S and V constant and varying H.

In general, the temporary markers would only last while the window is open, so all it would take would be a bit of extra code in the widget library — no changes required in the actual applications.

[permalink] ‣ keyword: idea

15 October 2006, 12:33 UTCIdea: commentary wiki

A wiki where some of the pages automatically include fixed text or images (that can't be changed, nor emulated on other pages). Perhaps this fixed part would be in a bar down the left hand side or something.

The wiki would then form commentary on the fixed text.

Implementation seems unproblematic - just modify the software to check a database of fixed parts when rendering pages. The only special case would be when a fixed part exists and the wiki page does not.


Related software:

[permalink] ‣ keyword: idea

13 October 2006, 8:01 UTCDistributed virtual worlds

On a mailing list I'm on, somebody wrote that virtual worlds based on open standards would be very cool but that finding CPU power and bandwidth to run them would be challenging. I responded, only a little flippantly, that the clients can do it.

The ultimate version of this would be a purely peer-to-peer system, with no central server at all.

A few problems / opportunities…


NAT is a pain, turning the simplest network program into a crawling horror. Still, standard NAT traversal techniques exist (with standard problems). [Update 12.3.2007: telepathy tubes might help here.]


Traditional currency requires a central server to act as the national reserve and the bank. However, other systems can be imagined: Cory Doctorow's Whuffie comes to mind, and its Wikipedia page links to Local Exchange Trading Systems and Ripple monetary system — the last being the monetary system for an open-source peer-to-peer distributed social network service; exactly what we need.


Second Life allocates virtual land area to players as one of its main sources of income. Any time there is a shared geography, somebody needs to be in charge of it. However, this may be an artificial scarcity.

For instance, players might have their own virtual rooms (or worlds), with portals to those of others — the VR equivalent of a friends list. These could be see-through, looking and working like doors (or see-only, like windows).


The system needs some mechanism for policing anti-social behaviour — either by banning users (the traditional solution) or by somehow mitigating their impact. However, perhaps a distributed trust system could be used here, too, perhaps a ban/approve weighted by how much one trusts those doing the ban/approval (or how much the admin of the room you're in trusts that ban/approval).

One might tie policing to the monetary system by charging in-game currency for anything that has the capacity to be harmful (so that bad actors are limited by their finances). I'm not sure whether that would be a good idea, though; it would certainly change the character of the game.

Update 5.11.2006: See also imvu.

[permalink] ‣ keywords: idea, ripple

29 May 2006, 5:20 UTCIdea: What to Do?

It'd be nice to have a gadget that randomly picks one thing from the various lists of what one should be doing... both permanent/semi-permanent to-do lists and the various issue trackers that one should be following, weighted by the priority (and the importance of each tracker).

Then late at night I could sit at it and hit "reload" repeatedly...

[permalink] ‣ keyword: idea

23 April 2006, 14:58 UTCOpen social sites?

I wonder if one of the problems of social sites would be that they're closed. Closed means there's no experimentation, which means there's little progress.

It's not very difficult to experiment in that space, but it's not easy, either. If you want to experiment, you have to start your own site, with the maintenance burden that entails, and then hope people actually join. That seems like a lot of effort just to experiment with a small little feature; and of course it'll take years before you have any idea whether the small little feature is actually any good, because that's how long it'll take to get any sizeable audience. Amassing an appreciable amount of progress that way will take decades.

Maybe what's needed is some sort of extensible social site; somewhere one could experiment with a small little feature by writing a little script, and have it immediately visible to the existing audience of that site. The challenge would be making it secure, but that would be doable.

Or maybe it already exists and I just don't know about it.

(previous entry)

Update 1.6.2006: I wonder if a small-sites-loosely-connected model would work.

[permalink] ‣ keyword: idea

21 April 2006, 14:18 UTCIdea: low self-esteem tee

BB recently linked a t-shirt which reads "hate" normally but "love" in a mirror.

I'm thinking maybe one could make a "low self-esteem tee", which reads "cute" (or something) normally but "ugly" in a mirror. I know a few people like that, who are cute really but see themselves as ugly in a mirror...

Edit: Perhaps a little like this (although perhaps a little neater and more legible; it's the first time I've ever done one of these):

Here it is in the mirror:

[permalink] ‣ keyword: idea

20 April 2006, 15:05 UTCIdea: security grille for leadlight windows

I've been musing on this idea for a while now...

Leadlight windows look nice, but they are not so good for security - the cames can be peeled off silently; or, of course, a brick can be used. Putting a security grille on them is effective, of course, but looks ugly - what's the point of leadlights if they're behind bars?

One solution might be to take a photo of the window and then laser or CNC a security grille which copies the cames. Where they are too far apart, the gaps would have to be filled in; still, with a bit of skill and care, it could easily be made to fit in with the style of the window.

Of course, for new windows, the cames might be lasered or CNC'd out of a harder metal to begin with, which would eliminate the problem altogether. However, it would certainly give a different feel to the traditional leadlights. (Admittedly, so does adding a security grille to them, no matter how well it may fit the style.)

[permalink] ‣ keyword: idea

31 March 2006, 5:13 UTCBruce Sterling on physical instantiation

interview - Bruce Sterling (via BB) I'm an author. I can remember when a typed manuscript of a novel was a desperately precious, idiosyncratic thing. Now it's a printout, an instantiation of an electronic word-process. Nobody's excited about printouts, not even fanatical book collectors. The value of a holograph ms has been radically altered. They're just not something you sweat blood about. Ever.

Imagine having that attitude about cars.

It will certainly be an interesting world... and it's not far off, either, there are such machines today (3D printers, realisers) - some working in papier-mâché, others in steel or titanium.

Somebody so should hook up one of these machines to the Internet, CaféPress-style.


[permalink] ‣ keywords: link, idea

29 March 2006, 14:33 UTCIdea: Pattern Paste

A variant on pfh's Smart Remove gimp plug-in, which would use the clipboard as the source rather than surrounding image.

Smart Remove is excellent for removing bits of an image surrounded by a uniform texture. However, if the object is near an edge, Smart Remove sometimes picks up bits from the other side of the line. If the object crosses an edge in the background, Smart Remove tends to be lost altogether, even if it's a very simple edge (for instance, a plain, flat horizon).

Pattern Paste would be used like this:

To remove objects crossing a line, one would do this separately on each side of the line (and then perhaps fiddle with the line itself by hand).

This could also be used to draw images, as a more flexible variant on Pattern Fill.

One of these days I'll have to learn Script-Fu and implement both this and Smart Paste...

[permalink] ‣ keyword: idea

20 March 2006, 5:44 UTCIdea: supermarket barcode reader in phone

Dad and I were thinking that it would be useful in the supermarket to have a gadget which reads barcodes and looks up product information - whether for simple reviews ("Jane's guide gives one star"), political activism ("produced by slave labour"), religion ("treif"), taste ("contains MSG") or even medical concerns ("contains MSG"). Particularly the sight-impaired allergic shopper would greatly benefit. However, the idea is hampered by the lack of appropriate barcode readers.

Many phones contain cameras, though, and some can be programmed. It should be reasonably easy to use the phone as a barcode reader. Any phone with a camera will have plenty of space for a database, and it already has both screen and sound output. If necessary, it could download data in real-time over GPRS. You'd probably have to put a macro adapter on the lens, but apart from that it should be fine.

Products already have some of this information on them, but it's annoying to read the tiny print, searching for the lack of half a dozen ingredients. A programmable gadget could be set to watch for those half a dozen ingredients and just come up with a summary according to my priorities.

[permalink] ‣ keyword: idea

18 March 2006, 14:46 UTCAlternate reality games and geocaching

All the extant ARGs seem to be puzzle-based, which means that they can only be played once, when they first appear - which also has the problem that the publisher must acquire a large audience before there's anything to show them and the game must be designed and written by a closed team working in secret.

Surely other types of games would work? Maybe some could be played repeatedly, in distinct rounds, or they could be just ongoing games. Surely internet-powered collaborative authoring would be more powerful than a closed team?

In some ways, geocaching probably counts as borderline ARG, and it is in fact quite the opposite in these respects: it's an ongoing game and it's collaboratively written.

I wonder what the confluence of these two will be; presumably it will be a game where writing and playing are commingled into "participation", so that each move is both a solution and a new puzzle, an answer and a new question.

But it hasn't been invented yet...

(17.5.2006) Rogaining might be another source of inspiration; in some ways rogaining and geocaching are variations on a theme, with rogaining being sharply limited in space and time whereas geocaching is unlimited in both; and rogaine is set by a small committee in secret while geocaching is distributed.

(21.5.2006) One way to grade questions posed might be as OKCupid does, by the degree to which they are informative. That is, a question nobody can answer is just as bad as one everybody can; ideally, you want a question that half the people can answer (cache half the people can find, puzzle half the people can solve, etc).

Of course, this needs to be modulated by difficulty levels; you want questions that are informative as far as beginners are concerned and you want questions that are informative when advanced players attempt them. But that, too, can be automatically calculated... Give each player whichever question they are estimated to have 50% chance of answering.

Actually, you might want to make that ⅔ chance or even ¾, rather than ½, for psychological reasons. Or marketing reasons. People like getting the right answers.

(21.5.2006 bis) Actually, the informativeness goes both ways: you want information about the difficulty of the question as well as information about the skill of the player. Fortunately the 50% chance of answering maximises both, other things being equal; which they aren't, but they're second-order effects like how many other players have answered the question versus how many other questions that player has answered. By and large, optimising both should work well.

(22.5.2006) In some ways, the ranking system becomes an epiphenomenon: the real goal of the system is to match up players and questions so that the questions present the appropriate level of challenge; the ranking system (if any) is subsidiary to that goal.

(29.9.2006) I wonder if something like snob could be productively used on the resulting data - perhaps to create guild-like groupings of players... On the one hand, it would group similar players together. On the other hand, though, the resulting groups have no overlap, which is not necessarily ideal from a social point of view (depending on the purpose to which these groups are put).

[permalink] ‣ keywords: creative, idea

16 March 2006, 9:55 UTCData storage crystals

On all the sci-fi shows (or, at least, Stargate) alien data storage devices are always crystals. So, why not make one's own?

Handsome clear resin casting for a broken USB drive (via BB)

He wasn't going for an "alien crystal" look, but even so the "night-time" shot at the bottom of the page looks quite close. To be more realistic (scifiistic?), one should probably make the resin block bigger, and put a bunch of LEDs around the base (or, alternately, build them into the hub) so as to have the whole length illuminated.

[permalink]1 comment ‣ keywords: idea, link

5 March 2006, 5:41 UTCIdea: Ghost Diagrams for text illumination

I wonder if a variant on pfh's ghost diagrams could be used to illuminate text...

Just add appropriate ports on the characters, and let it doodle. It would have to be continuous rather than grid-based, and it would probably need to have some flexibility in the pieces so they can join up. It should definitely get the ct and st ligatures, and if you make the loops on various letters flexible, you'll also get ff and fi and of course ties between adjacent lines, where the descender of one letter is joined to the ascender of one below.

After all, the future is textured :-)

[permalink] ‣ keyword: idea

3 March 2006, 11:29 UTCIdea: program to teach approximate calculation

One skill that schools don't seem to teach much, but which would be very useful, is approximate mental arithmetic. The kind that says that 28 divided by 5 is a bit under 6.

Surely, though, a program could be written to do that? One that rewards quick, approximately right answers?

[permalink] ‣ keyword: idea

27 February 2006, 9:18 UTCIdea: flickrwar

Like the voting on cute overload or kittenwar, but based off flickr tags.

[permalink] ‣ keyword: idea

8 February 2006, 3:08 UTCIdea: fruit market

Dad and I were grumbling about the quality of fruit these days - it's bred, grown and harvested for looks and transport, but no regard for taste and texture, with the predictable result that taste and texture suffer. A lot.

Dad suggested that perhaps the Internet should be able to fix this; after all, if the supply chain is shortened to a single hop - farmer directly to foodie - then transport issues aren't a problem, especially if the foodie comes to pick it up. Even overnighting should be OK, most fruit will be able to handle that.

Shortening supply chains - disintermediation - is the Internet's specialty :-)

Of course, we could grow our own, but we lack the requisite green thumbs, and occasionally going away for several weeks doesn't help. Besides, one tree means one kind of fruit, and a whole orchard is too much to eat... We'd be happy to pay someone else to do the work, though.

One way to organise this would be with eBay; it's not a perfect fit, but it has the virtue that it's already there - all you have to do is start using it. No doubt people already grow such fruit for their own use, so starting the system up would just consist of starting to trade the excess...

As eBay rightly points out, selling food is regulated; however, I do believe selling fresh fruit in "garden excess" quantities should be reasonably OK (anybody know details?). Overnighting fresh fruit would be subject to pest control regulations (fruit fly zones), but I don't think the direction from country to city is much of a problem for any of those... Anyway, would have to check.

[permalink] ‣ keyword: idea

25 January 2006, 4:05 UTCIdea: wiki with bug-tracking

Bugzilla is good at what it does, but it feels restrictive. You have to do things its way, and it's in control rather than you. This gets more frustrating the further your project is from Mozilla.

One possibility would be to add bug-tracking capability to a wiki rather than the other way around. Anywhere in the text, you might mark-up part of the text as a bug title, optionally with things such as keywords. Perhaps this would display as a graphic... This would mean, for instance, that the person writing a HOWTO in the wiki can put a bug in-line asking for the procedure to be simplified.

The software would provide lists of open bugs, but there would be no "Description" field: clicking on the bug would take you to the wiki page, which presumably contains a sufficient description of the problem in the context in which it was made.

Possible syntax: !! title of bug (keywords, options) !! with the number of exclamation marks indicating priority :-)

The software might or might not have facilities for automatic changes of bugs. In some ways, it's not very important, since one will usually wish to change the surrounding text anyway; but it would be a good feature to have.

To some extent, wikipedia templates such as {{cleanup}} are a simple version of this; they concentrate on the contents of the wiki itself rather than whatever it's describing, but that's not such a big difference. Probably the biggest lack is that they don't (generally) allow for bug titles and/or other additional info, so that the summary pages are necessarily restricted to just a listing of pages for each bug category.

Bug history is subsumed by the wiki's "history of this page" function.

[permalink] ‣ keyword: idea

8 January 2006, 10:04 UTCIdea: Voluntary Taxation

So, if the RSPP makes it "rational to contribute money towards the production of the public work", it should work on a grander scale as a replacement for the taxation system, no?


[permalink] ‣ keyword: idea

8 January 2006, 9:29 UTCIdea: Automatic SingStar

Sony's SingStar is nice, but it's limited to the music specifically prepared for it, which is rather limited. Even with all the expansions, it's not even a gross, and it's a lucky artist indeed that has more than one track for it.

Surely, though, it ought to be possible to automatically score on the consonance or dissonance between the player and the backing?

At that point, any karaoke track would be available, which is an immeasurably larger selection. If one is willing to forgo the lyric prompts, one can use any track at all. Either way, you can sing all of your favourite artist's songs, not just the single one Sony have picked.

[permalink] ‣ keyword: idea

28 December 2005, 14:00 UTCIdea: the blind matchmaker

Idea: A matchmaking site which sets people up on blind dates. Sort of a cross between dating site and All the site would need to do is match up members (who have time this week) in pairs, based on the kinds of measures a matchmaking site would use, match them up with venues (according to geography and preferences) and send out a bunch of notices based on the result...

[permalink] ‣ keyword: idea

17 December 2005, 3:21 UTCIdea: flicker-free light

I'm surprised I haven't seen anybody selling these...

Electric lights running off the mains flicker - fluoros more, lightbulbs less, but they all do flicker. Some people get headaches in proportion to the flicker. Yet it should not be a problem to put together a light that does not flicker at all...

All it'd take would be a battery, smart-charger, a headlight off a car/truck/locomotive and a big switch. Maybe a battery-guard to stop the battery getting ruined when someone forgets the light on.

[permalink] ‣ keyword: idea

15 December 2005, 15:40 UTCIdea: faux mayan long count clock

A day is 12 × 20 × 18 × 20 seconds. One could make a faux mayan long count clock... with four quasi-vigesimal digits, like this: two-hours . six-minutes . minute-thirds . seconds

These aren't even such bad time-keeping units, especially the minute-third and the six-minute. PS: And the two-hour is an old Chinese unit of time...

[permalink] ‣ keyword: idea

24 November 2005, 2:57 UTCIdea: reputation vs quiz memes

I wonder to what extent it would be possible / good to model a reputation system on quiz memes. Instead of a blank "testimonial" box (as on orkut), or a one-dimensional "+/-/neutral" rating (as on eBay), it would be a quiz meme. The only difference would be that it would be phrased about a friend rather than yourself.

"If <friend> were a Star Wars character, he would be Luke Skywalker."

It would be really easy to do, too - there's plenty of sites providing quizzes, and you can just put the results as a comment in your friend's LJ. That way, it's clearly associated with both of you, signed or anonymous as desired, and your friend has the option of screening / deleting unwelcome testimonials.

Completely doable today, all it'd take would be someone putting up a quiz like that.

(next entry)

[permalink] ‣ keyword: idea

16 November 2005, 9:39 UTCIdea: wiki game gameplay - items carried

Recently, a friend launched a wiki game called Deiludum. I've also come across an article about Destiny of Online Games, which seems to be pointing in the same direction, toward user-built games with a "small pieces loosely joined" philosophy.

Here's an idea for gameplay: the inventory of items carried could be implemented as cookies. There would be special elements in the wiki that would add (or take away) items carried ("pick up" and "use up" links), and also elements to present different text if the player has particular items.

Most likely, an item would be represented as its own name; thus, there would be elements like [pickup rusty key] and [if [have rusty key] [link Open door with rusty key [page 57]]] (or whatever syntax the wiki uses).

The wiki would also automatically generate a "You are carrying" box with links to the items, perhaps prefixed by Item: or something.

PS: It's been pointed out to me that, of course, using cookies for game state is nothing new. The new element here is the combination with a wiki, allowing web-based development of the game.

[permalink] ‣ keyword: idea

28 October 2005, 18:28 UTCIdea: voted wiki

Like wiki, except instead of editing being amendment, editing would mean proposing a new version. Registered users would vote on which version should be "official".

(next post)

[permalink] ‣ keywords: vote, idea

15 June 2005, 4:00 UTCIdea: streamtorrent

Why not stream bittorrent? For people who are exactly on-the-minute it probably wouldn't be so useful, but for people who are watching five minutes later (tivo-style), it should be just the thing...

Some modification to the protocol(s) would be required, but it shouldn't be all that difficult.

(added 9.10.2005) Similarly, it would be good for people who send each other links to particular points in a show, or comment about them on their blogs. To that end, the URLs should probably allow per-frame timestamps, either from the beginning of the show (for films or episodes) or the wall-clock time (for live streaming video or real-world events).

Especially for live streaming video, this would be a powerful model. The "reality TV" show streams a dozen cameras, then people post links to the particularly juicy bits. So does the producer, of course, but a lot of publicity can come from people forwarding links around...

(Hmm, the client might even support scene lists, where it takes a list of timestamp ranges from various sources and cuts them together. Not sure how good this option would be netiquette-wise.)

The protocol would help people watching live, too. As the server gets overloaded, people fall behind and start to get the show off peers, therefore lightening the load on the server. Graceful degradation.

(added 9.11.2005) See also Boing Boing's BBC Radio's experimental audio-tagging project, linking to Public could help BBC to index archive.

[permalink] ‣ keyword: idea

28 May 2005, 9:32 UTCIdea: rtorrent

The rsync protocol speeds up file transfers by taking advantage of the recipient already having an older version of the same file.

The bittorrent protocol speeds up file transfers by taking advantage of other people downloading the same file at the same time.

Surely these two should be combined?

When a client has a torrent and a candidate older version of the file, it should first search the older file for the pieces (at possibly different offsets, rsync-style), before asking for them from the network...

[permalink] ‣ keyword: idea

28 May 2005, 7:53 UTCWishlist: graph layout widget

It would be nice to have a widget which does graph layout - nodes, edges, that sort of thing. Both node placement and especially edge routing should be automatic.


In the same way that the Tree widget facilitates display of tree-shaped data, and makes it thereby common, so a Graph widget would facilitate display of data best suited to that, thus extending the kinds of data that can be easily displayed (and therefore easily manipulated).

"Civilisation advances by extending the number of operations we can perform without thinking about them." (Alfred North Whitehead)

This is not intended as a graph editor; it's intended as a browser or editor of other data, which happens to be displayable as a graph.


Nodes would have a gtk.Table widget in them, and edges could have optional restrictions about which ends of which rows/columns of a node they may attach to. That way, named ports are possible. Edges would also have widgets, one for each end and one in the middle, but probably ordinary ones (usually labels, likely).

The edges might be a bit generalised, so that they are sets of nodes rather than pairs; that way, one can have singleton edges, which are displayed as visible/clickable ports, and multi-way edges, for when several things need to be connected but the order doesn't particularly matter (as in circuits). Each end of an edge has separate settings for arrowhead or not, anyway.


[permalink] ‣ keyword: idea

27 May 2005, 13:59 UTCIdea: mat2mat protocol

Idea for a module for the MatPLC:

The module would contact another copy of itself running on another MatPLC, and automatically configure exchange of data between same-named points.

Any point configured for writing would requested from the other end of the connection, while any point configured for reading would be offered to the other end. Any mismatches (writeable points not available from the other end, differing point sizes) would be logged as errors.

This means that the configuration is strict yet light - the best kind...

There would also be a stand-alone program which contacts the mat2mat module and asks it for all the points it's offering, displaying them on the screen in some interactive format.

Come to think of it, this last would be good locally, too (directly connecting to the MatPLC rather than over a network).

[permalink] ‣ keyword: idea

3 December 2004, 9:49 UTCIdea: article ratings for a blog-reader

Idea: a cross between circle's article ratings and a blog-reader (RSS/Atom).

As with a normal blog-reader, the user would subscribe to a bunch of RSS and Atom feeds. Each article would have ratings buttons (cool/uncool, timely/old). In addition, the blog-reader would present articles you aren't subscribed to, but which other people of similar tastes rate highly (either manually-configured friends, or automatically-detected based on past ratings).

(update 7.2.2005) More thoughts on this idea in a later entry. (update 6.4.2005) Including links to existing software...

[permalink] ‣ keyword: idea

24 November 2004, 5:03 UTCClassless Charity Slave Auctions

In a traditional charity slave auction, the participants are first divided into slaves and buyers. Could this distinction be eliminated? Here's one possibility...

The participants are all in one group. Each participant bids on several others; the bids are collected and a combination calculated so that everyone is involved in at most one successful bid (whether as master or slave), while the total donation is maximised. The solution is displayed, and participants can up their bids until some pre-determined deadline or total.

Google suggests that a computer can calculate such a combination easily - here's the code, complexity O(n³) so it should run reasonably quickly. It's for undirected graphs, but that's OK - mutual bids are resolved in favour of the higher amount, which leaves a plain graph to work with.


Implemented! (new 19.10.2006) Download the code.

(You need both this and the weighted matching code.)


[permalink] ‣ keywords: idea, script