Stuffz

Idea: Wireless tablet interface

The idea is to have a tablet size touchscreen that has no cpu but rather connects wireless to any computer via either something like VNC or using some kind of special usb stick that you plug into the target computer. The advantage of the VNC method is no physical modification to the computer is required. The advantage of the usb stick is no software modification is required (potentially).

The point behind this is to make any desktop computer portable within a short distance of the physical base. This would free computers up from needing to be lugged around, aka a dumb terminal touchscreen tablet.

I suppose the VNC method combined with internet connectivity would be ideal because then you could connect to any computer on the internet by joining a wireless network, you’d just need enough smarts on the tablet to have some kind of firmware interface for joining wireless networks and talking through them. Add in 3G or other wireless broadband (or at least the ability to tether with a phone) and you suddenly have an ultra portable with a large screen that can switch computers.

Idea: Tag browsing hierarchies

Problem with Tags: Tags are flat browsing systems, making specificity a problem. Tags need to be broad and there’s a limit to the depth of specificity with combinations of broad tags resulting in lots of manual browsing.

Problem with Hierarchy trees: Hierarchy trees work like tags that get more specific as you get deeper into the tree. The only way to provide multiple tags is in your branches which can lead to lots of redunancy issues. Hierarchy trees are bare often either drag and drop, or single clipboard cut and paste, making tag manipulation wieldly.

Solution: To get the speed of a flat browsing structure, combined with an element of tag nesting, I propose a modified tag cloud system. The way a normal tag cloud works is  tags that have been applied more frequently are given a bigger size. While the proposed system may actually use a different indicator than size (such as list ranking), its based off the same concept of frequency indication. The modification to the tag cloud idea is to filter the tag list based off selected tags.

The way this would work is the most broadly applied tags (those being “higher level” in a hierarchical sense) would be given prominence in the tag list. Then when any of those tags are clicked, the tag list would be culled down to only tags that have been used in combination with this tag. When another tag is selected, the list would be culled further in the same manner. This provides a hierarchical drilling mechanism to give the benifits of a hierarchical file structure. However because the list is still flat, you can skip levels of navigation quickly. The skipping will be easier once you’ve culled the list to a certain amount of course, depending on the density of your tag cloud.

Let me provide an example:

Let’s say you want to find a specific video you have been working on for a client (Microsoft), and here is your starting, unfiltered tag list with the broader tags first.

Personal, Work, Documents, Images, Video, Sound, School, Clients, Math, Science, English, Ads, Apple, Microsoft, MacvsPC, Switch, SeinfeldandGates, ImaPC, Porn

Now to start out with, thats a lot of mixing of different items ranked solely by frequency. It may look a little wieldy at first, but with a tag list this short we can simply click on “SeinfeldandGates” and “Video” to get right to the list of videos to look through and find our file quickly. However, let’s pretend that list is exteremely long and looking through it to shortcut so easily wouldn’t be as easy and look at the longest approach.

1) Clicking on Work culls the list down to: [Work], Documents, Images, Video, Sound, Clients, Ads, Apple, Microsoft, MacvsPC, Switch, SeinfeldandGates, ImaPC

2) Clicking on Video doesn’t cull the tag list down but we are going to want to do it at some point anyways because it culls down the results list.

3) Clicking on Clients and ads would reduce the list down if have any work tags that aren’t related to clients or any client tags that aren’t for ads.

4) Clicking on Microsoft however will drop this tag list down to the following: [Work], [Video], [Clients], [Ads], [Microsoft], SeinfeldandGates, ImaPC

5) Clicking on either of the two ad campaign tags left then will give you a list of the videos for that campaign, enabling you to find the one you want. Of course you could have some tags such as “Drafts”, “Unedited”, and “Finalized” to narrow that list down more.

That was an example of taking the longest approach, realistically all that was needed was clicking [SeinfeldandGates] and [Video], the rest of the tag clicking was mostly to narrow the tag list. This provides tags with two functions, thinning the tag cloud and thinning the tag results. An example with a much larger tag cloud would obviously be more dramatic but i’ll leave that to your imagination.

The one weakness in this system is the dependancy on making sure you tag more specific items with both the specific tags AND the broader tags. How detailed you want to be in this though is up to you since you are free to skip over intermediary tags in the tag cloud at any time. If you had a tag cloud as small as the example above, you could merely tag all your SeinfeldandGates videos with that tag and then skip straight to that tag and browse for the video files manually. This means density is up to you :)

Also as an additional note, because this tag list technically allows for unlimited specificity, it could be imported from an existing file hiarchy and then the user could modify down the tag list himself if the import didn’t turn out well.

Idea: Graphiti

Most graffiti that is respected is either artistic or trying to make a statement. When trying to make a statement, how about saying more than just “Bobo was here”.

Why not get data that supports your statement, say number of civilian deaths in iraq and make a simplisitic but visually interesting data visualization such as a graph or plot. Then make that into a stencil of sorts for easy spray graffiti-ing. That way you can spread facts in the public forum instead of just opinions. The uninformed can always use more useful data! Also I liked the play on words of the title so stfu :)

Idea: Friggin Room Bass Cancelation

nomel:

noise cancel stomping bass hit from the kids that live above us EVERY FRIGGIN MORNING. Woofer placement would probably be critical. Could use something like digital room correction by the awsome Dr. Denis Sbragion to optimize the room response near where your head sleeps (disjointed bodies are ok).

If you could actually make a commercial product of this and get it installed in apartments everywhere you’d be rich!

Idea: Finding parts on a PCB.

nomel:

Problem: On most PCB’s, you’ll find hundreds of parts. When debugging/testing/modifying these boards, you often have to look at the schematic, then find a part on the board. With potentially hundreds of parts on the front and back of the board with tiny little stencils, this can burn up a lot of time and fun.

Solution: An automatic part finder than uses an inexpensive video camera, a laser pointer, and a laser pointer positioner (probably servo controlled).

The software will find the board in the frame using feature and color detection, then board orientation and the visible side (back or front) will be found by finding known features on the boards surface. Once orientation is known, the laser positioner will point the laser to the point to the part on the board, known from the layout file of the board. Since this can be done in realtime, the lasers position could track the part, even on a moving board.

Use/Features: Enter a part number to find the part. If the part is located on the back, the laser will blink. If you wanted to get real fancy (for EMI debugging, etc) you could have the laser follow the paths of the traces between two parts.

Cool idea! Could definitely have uses in other areas besides PCB boards, any kind of cartography related activity in general really. This is basically just lower tech (because the laser pointer is the only display used) augmented reality. Luckily optical tracking stuff has gotten really good. You could probably pretty easily orient the board for the optical sensor using colored dots placed on the corners of the board or something similar.

Tagged email addresses to fight spam

Been thinking this morning about the problem of fighting spam and it seems to me that gmail has one of the best tricks around at the moment which is to allow you to use the format <yourusername>+<anythingyouwant>@gmail.com. This means that you can generate unique e-mail addresses anytime you want.

Sure you could always sign up for websites with something like mzomzo+spambox@gmail.com and then use filters on anything going to that e-mail to shove it to the spam box but that seems like a bit of a shotgun approach still since you are likely to kill false positives.

Why not use a unique tag based on who you give it to, like lets say you sign up for playboy.com, why not use mzomzo+playboy@gmail.com. This way if you get any spam to that address you know what the source is and you can block any spam coming from them selling off your e-mail from now on using a filter.

Extending that idea you could go really crazy and instead of posting your normal e-mail address on your website you could have your website generate a random or incrementing unique e-mail address everytime the page is viewed so anytime your e-mail got harvested you could block any e-mails going to that address knowing full well that only 1 person (plus anyone that person sold it to) is using that address. The downside of this of course is that you can only setup so many filters in gmail so perhaps this is an idea for a custom e-mail service that could easily block any future spam from an address with a single click (like google’s Report Spam). Also you wanted to use it outside of your website you could simply make up the random/incremental part yourself or use some kind of widget or token to generate it.

Seems to me the only way spammers could detect this is if recognized that anything after + in a gmail address was bunk and just ignored it. I suppose this is where a service that generates full one off e-mail addresses would be useful. As far as I know none of them generate lots of one-offs linked to a single account easily.

I see it working like this:

1) Sign up for an account on the webmail service.

2) Get some kind of widget or whatever portable e-mail addy generator.

3) Generate e-mails automatically linked to your account.

4) Click a “SPAMMER” button on any e-mail spam to completely remove that randomly generated e-mail address from the system. That means no more e-mails to that e-mail address will come to your box and it means that the webmail service can purge that account from its system to save space, perhaps marking that e-mail on a deadlist so it never uses it again.

Wonder how hard this would be to throw together with Django and the Google App Engine using the Mail API…

Food for thought.