tsunami

log in
history

old tsunami main page

Luke Breuer
2010-05-20 18:04 UTC

introduction
Five years from now, tsunami will be a portal to all of the information that you want & need. It will let you connect disparate types of personal data as well as provide you with a customized and dynamic textbook for anything you want to learn. You will be able to query, slice, and edit information fluidly. You will be able to discuss any piece of information via any type of communication, and then be able to look at a piece of information and see where it was discussed.
a portal to your data
Currently, you probably use multiple programs on your computer, or at least multiple web sites, to manage all your information. Sometimes this is essential: you need an email client to send email and many out there integrate lots of functionality that you want to have present when emailing people. However, sometimes this is really undesirable: if you want to search for something, you may want to really search across all data you care about, or at least more data than one program stores.

I want tsunami to contain some of your data and be able to access the rest. This way, you will be able to search all your data from one place. However, that's not all: if you want different pieces of data to refer to each other, with relationships that a computer can understand, then you're stuck without tsunami. Most email clients have no way to say that two items are closely related, other than to stick them in a folder by themselves. Never mind relating an email to a particular URL (unless you edit the email). Tsunami will let you relate items together, optionally let you specify why they are related, and then let you use these relationships when you are searching for, displaying, or aggregating your personal data.
a customized, dynamic textbook
Consider a given textbook on some subject. Here are some things I do not like about it, without knowing anything more:
  • once the textbook is printed, corrections must be made in subsequent editions (which must be bought), or errata, which must be manually correlated to the associated erroneous text
  • questions asked of the textbook are usually not associated with it, making it hard to improve it on the fly or even see which parts are unclear
  • the textbook costs a lot of money
  • it takes time to access the contents of the textbook even by library so there is no monetary cost
  • it takes more than {the time to click on a hyperlink} to view referenced material
  • it might be nigh impossible to see what references the textbook
  • the textbook is written toward an audience with some set level of knowledge with the subject, making it hard for those without the proper background and annoying for those who know most of the material

I want to make electronic textbooks which are not susceptible to the above flaws. In particular:
  • it should be possible to reshape content (perhaps by Adaptive Content) based on what the reader does and does not know
    • "make this section more detailed"
    • "I know this concept"
    • "give me more examples"
    • "please help me with this kind of math"
  • it should be easy to ask a question, get an answer, and then have both closely associated with the relevant section(s) of the text
    • these can then be addressed "permanently" by modifying the text appropriately
  • sections in the text can symbolically link to sections in the same or other texts
querying, slicing, and editing
Sometimes you just want to find all pieces of data that match some search string. However, sometimes you want to exploit special aspects of your data, like:
  • length of the MP3
  • addressee of an email
  • date range of an appointment or todo item
  • access rights of some set of data (like all public calendar items)
Searching for information according to the above data is either impossible for most users, or each requires using a custom search screen that is usually convoluted. There is no technical limitation that results in this! Instead, it is the refusal of software companies to expose standardized querying interfaces to their data that results in this problem. By accessing your data through the tsunami portal, you will be able to search for "filter by mp3 length" and be provided with a simple interface for conducting your search.

Different data clearly need different methods of presentation. Some data can easily be aggregated (by month, for example), and often you only want to look at certain characteristics of your data (like summary). Again, different pieces of software have different ways of doing this or no way at all. The tsunami portal would make this trivial.

Not only will information be viewable from the portal, but it will be editable as well. You might even have a single screen that shows an email draft, a wiki article, a few todo items, and relevant contact info; all will be editable from that page!
better discussion of ideas
I believe that the main, current methods of discussing ideas online are subpar.


V archived V
Tsunami allows people to collaborate and create high quality information with as little effort as possible. Tsunami "gets out of the way" so that you can communicate exactly as you want. You can switch from email to chat to a wiki; whatever you use, tsunami will help you collect and connect the information being exchanged. Then, tsunami will allow you to organize and transform it into a succinct, cohesive whole. Whenever new information comes in, it can be quickly connected to the existing information. The end result is knowledge interrelated in ways that has never been seen before.

You almost definitely read more material than you write. Taking all the readers and writers in the world into account, I would probably estimate that people read 1,000-100,000x as much as they write. What if we ask the writers to take more time in what they write so that the readers can read more quickly?

Note that paragraph 1 is an attempt to minimize the amount of energy required for paragraph 2 to hit critical mass.
a story
I needed to use a context menu in javascript one day and looked for good online references. I found unwieldy, overly verbose tutorials with sub-par code that only covered a fraction of what I needed. As if to put salt in the wound, these sites invariably required me to visit multiple pages of material laden with advertisements, cruft, and debris. Some wanted me to register in order to see anything.

I set out to fix this by writing my own tutorial so others would not have to suffer similarly. Search the web for "javascript context menu". My webpage is the top result or close to the top result on google, yahoo, and live. This is no coincidence. Due to this popularity, I wrote a few more tutorials. You can see the result (popular searches and bcp searches). I get an email or two a month thanking me for my work, and three people have actually donated $10 each. (I think that works out to about $1.50/hr.)

I can only churn out so much high quality content per unit time. My goal is to make a web platform that enables others to do the same, with the help of a community that critiques and enhances the content. Wikis are supposed to support this, but they have many problems, like having no good discussion mechanism. With some information-centric forums intricately linked to a wiki designed to foster high quality work, I hope I can draw more people to my "cause".
dead trees
The biggest problem with textbooks is that they are immutable. The biggest benefit is that they are mutable. Here is how I am defining those terms:
immutable — errors in the text cannot be corrected until the next edition
mutable — it is easy to mark up the text in a variety of ways

The goal of tsunami is to get the benefit of both kinds of mutability:
  1. ability to update the content in real-time
  2. ability to mark up the text in [mostly] arbitrary ways

My hope is that #1 will keep articles in tsunami from ever becoming dead. This is the underlying principle of Wikipedia and it works quite well. One of my biggest problems with Wikipedia is that it is hard to mark up information and it is hard to discuss it.

The principal method of doing #2 is symbolic quoting. Once we have a distinct set of characters to highlight, we can choose many ways to highlight. All of a sudden, we can do things that cannot be done with books, like searching all highlighted text (or, say, text within a few words of highlighted text). When you are looking for something, the instant you can say, "I highlighted that!", your search space reduces drastically.
a symbiotic relationship
Almost every information system in existence does one of the below; it is almost as if the better it does at one thing, the worse it does at the other:
  1. store information
  2. discuss information

Consider the epitome of #1: portable document format (PDF). It can be used to present information in any form desirable, but extracting information out or updating it without the original source can be nigh impossible. In fact, PDF appears to have been developed to prevent this very thing.

The epitome of #2 is people in front of a whiteboard. The interactivity is at a maximum, but much is lost without recording the entire thing with a video camera. Even then, the utility of such information would be quite low because of inherent unsearchability.

The goal of tsunami is to get the best of both worlds and let the user choose which aspect to emphasize. My vision is that high quality content will "bubble up" the more it is discussed/enhanced.