What?
Oh, come on. That's okay.
Okay, what I'd like to do is I'd like to get the slackers up here, up into this portion of the
chairs.
I have got some t-shirts that I'm going to give away. We're going to do a Q&A session and hopefully we're going to have a little bit
of fun with this. Everybody looks bored. It's not going to be as bad as you think.
Here is the mandatory
personal anecdote. Okay, I'm in New Jersey right now.
Yeah.
So I'm on the plane, right? And we're getting ready to take off.
And I've got two guys behind me.
And I've got the cute girl next to him. And there's a guy over here that looks kind of goofy.
And there's another girl next to him, right? So we're waiting to take off and I'm thinking
these two don't know each other. These two don't know each other. We're headed to Vegas.
There's all kinds of possibilities here, right?
Then the dumb guy gets up, taps the girl on the shoulder and he says,
Hey, would you mind moving over so that I can sit with my friend?
That's when I knew that I had computer geeks behind me.
True story.
And somewhere I was hoping that there would be a guy here whose face would turn really red.
Anyway, so they're behind me and they start talking geek speak.
They start talking about XML and security and all this stuff,
which past a certain point you get kind of tired of it.
Then I got a surprise. These guys are all in their 20s. They're clean cut.
They seem really driven by what they're doing, which is typical for DEF CON, I think.
And then they're going through resumes and the one guy looks at a resume and he says,
This guy's got a lot of technical knowledge. And the other guy looks and he says,
Yeah, I interviewed him, but he doesn't have any business knowledge and he doesn't know
what is of value and what is not. And that surprised me.
That's what I look for now. I look for what things have value. I think we're in the midst of a change
in the industry where we focus on what is of value
rather than the mechanics of building things.
So that's what our focus is going to be. It's going to be on keeping people employed through the industry
because I personally don't believe that the shakeout is over yet.
This is my other personal anecdote,
which I decided was too egotistical. Why should you listen to me?
I don't have a black t-shirt on, so maybe you shouldn't.
What we're going to do, we're going to go through and we're going to hit three basic areas.
We're going to lay out a couple simple business concepts, which are very
common in MBA programs. They're common in
large companies. We're going to lay those theories out.
We're going to drive them down into a technical implementation
that gives us concrete results that we can make decisions from.
And then towards the end of the presentation, what I'm going to do is I'm going to throw out some strategic concepts
to you, which are going to be
a little bit radical, I think. We'll test it. We'll see. This is DEF CON. Maybe it won't be radical.
Maybe nobody will care. But there's some ideas that I think that people can play around with and they
might be able to turn them into additional concrete implementations
in companies, things like that.
Okay. Everybody watches the news, right?
I took this a month ago and this is already out of date because there's already like a dozen more I could add on here.
Layouts are still going on. People are still being laid off left and right.
The crash is not over yet. Historically, industrial
revolutions like this have seen a shakeout of approximately
30% of the peak population of the leading technologies.
What we've seen in IT right now is about a 15% shakeout. So personally, I think
that you're looking at another wave that will be equivalent to
the last four years. How many people here have been laid off or out of work in the last
five years? Yeah. And you know what? For every hand
that's here, there's a guy that isn't here because he's not in the industry anymore. He got washed out.
My goal here is to give people a little bit of knowledge
so that they can make decisions that will keep them employed or
it will let them make a decision to leave the industry now before they waste a lot
of time on it. So either way, if that's the outcome
here, I'm happy with either outcome.
The time and effort that you put out to learn things is finite. If you learn something that has
no value, you're fucked. Right? Your time is gone. You can't
get it back. People are not going to pay you for it.
Now, the skew here in DEF CON is a little bit different because you guys are in the security
industry and I think you've got a substantially better chance
of staying employed over the next four or five years than business.
If you're in business software, I think you're going to be in trouble.
IT market changes constantly, everybody knows that. What we're going to look at is we're going to try and focus
on what kind of skills that we should be learning to stay employed.
How many people know what a meme is?
That's good. The first question I got asked when I walked in was, what is a meme?
And I thought, oh my god, I should have put that up. A meme is an idea
it's a mind virus that passes through a population.
And typically, if you want to be completely theoretical about it,
it starts with a single person, that person talks to two more, those two people talk to
four people, four people talk to eight, so it's typically an exponential function.
And it looks like an exponential function, a
hyperbolic curve, until it reaches about the 50% mark of reaching
the entire population and then, because you're getting people that already know it,
you're talking to people, oh yeah, I already know that, I already know that, it starts tapering down and you get
a drop-off as you reach market saturation.
And that's typically known as an S-curve. How many people know the concept of an S-curve?
Okay, that's good. Here's a picture of it.
I start with a single point, it goes through a rapid run-up,
it reaches the 50% point, and then I reach market saturation.
This is a fundamental concept that you can apply in a lot
of different situations, a surprising number. And it's simple, it's a simple concept.
Ultimately, this is our goal.
We want to eliminate getting into new
technologies when they're first introduced because there's a very high failure rate. I don't want to be
here. I don't want to be in the high-risk area. I don't want to be in the low-reward
area because that's a technology on its way out and I can't make any money in it.
What I want to do is I want to identify
an area as it's ramping up, but I want to get it before
it's at the 50% mark so I can make the most money off of it. Right here,
I can make a decision that, yeah, this is probably going to stick around, this is probably going to
be a viable technology, and I can ride the highest part
of the wave off, which is the optimum profit point because other people are not paying
attention and you're already in the market to learn these things. You're already in the market to sell your services
in this. This is a simple concept. Does everybody grasp this?
Can I get a show of hands where everybody thinks that this is... do you think it's important?
The reason I care about an S-curve is it
has predictive power into the future.
How many people here know calculus? Oh, there you go.
So you know this already. As this is ramping up, my rate of
change is constantly increasing and I can measure that by doing empirical analysis
on certain things. And I can see that the saturation is...
or the propagation of the mean is rising.
Once it reaches the 50% mark, where roughly I have 50% of the people know
the mean's already propagated, it changes, I reach the inflection point
and the rate of change begins decreasing. So if I can measure the rate of change and see that it's
increasing, I know that I'm still on the pre-50% mark.
Afterwards, I know that I'm after the 50% mark. So this is really what we look for.
We look for a rate of change.
Okay, peak oil. How many know about peak oil?
Oh my god, now I'm
stunned.
Just as a quick overview, peak oil is the idea that
oil resources are finite in the world. And there was a guy named Hubert
which, in the 50s, he did basically an S-curve analysis
at the rate at which they were finding oil drilling holes, you know, that they were making
discoveries. And he graphed it down on a chart and he was able to see that in the United States
that peak oil production would occur around 1970. And he
was right within about a year. And he predicted it 15 to 20 years before it happened.
This is all numbers off the top of my head, so I didn't actually expect to talk about this.
What is probably very likely
going to happen over the next few years is you're going to see peak oil occur worldwide.
There's a lot of talk about it. If you do a search on the internet, Google peak oil, you'll get all kinds of
information. Environmentalists have been talking about it for a long time.
I'm not an environmentalist, but I understand the concept
of how Hubert did the analysis, and I personally believe that this is very likely to happen.
I think you're going to see peak oil happen within the next five years. And that's why
you see oil prices being driven up. It's because there's a lot of guys like me sitting at companies like
J.P. Morgan, and they're doing the same analysis, and they're saying, peak oil
is probably going to happen. We need to get our money into oil now and make money on that.
Oil prices at $60, right? Everybody thinks it's speculation?
I don't think so. This is why you care.
You can make money. Or you can avoid losing money, which
depends on if you're risk averse or not.
This is another example of an S-curve. How many people know Harry Dent?
Yeah, that's not surprising. Harry Dent wrote a book in 1989
which was called The Great Boom Ahead, where he predicted that there would be a rapid
IT boom, which would essentially look like an S-curve, and that's exactly what we saw
from 1990 to 2000. We saw the entire information
technology sector boom hugely beyond what anybody imagined, except Harry Dent
did predict it. I don't buy a lot of his other theories. I do believe he was right.
I mean, you can't deny that he was right about what he claimed. He made a second prediction
in his book, which is that there will be
a subsequent wave of S-curve technologies, which will build upon the
infrastructure that was created in the first S-curve. That's where we're at right now.
There's an infrastructure in place. There's networks everywhere. There's wireless
networks, there's standards, there's TCP, IP, all the acronyms,
all those things. It's too hard for me to remember. But ultimately, if you
want to be looking where you should be, you don't want to be
here anymore. This is foundational
stuff that's in place, that's standard. It's not interesting to work on. It's
maintenance work, and it doesn't pay a lot of money, and that's why you see a lot of things being outsourced now.
It's stuff that's duplicatable, that we can send somewhere where there's a different kind of culture
that adapts to those kind of things. What you want to be looking for is
technologies that build on the existing infrastructure that was
created in the last ten years.
And I goofed up because I was going to give away a t-shirt if somebody told me what peak oil
was and I blew it.
Okay, MemMiner.
This is a little tool that I made, which I've been doing this for about
thirteen years, and basically it assumes that a frequency count, a word count
on the internet is roughly equivalent to an S-curve growth of that technology.
That's a big assumption, I know. And how many math people do we have in here?
Oh, thank God. I don't have to worry about all the criticism.
You get a ton of criticism. From my standpoint, I'm very pragmatic.
I don't want to sit down and do a lot of analysis about probability and this and that. I just want a tool,
a rough cut tool that will give me a way to make a decision quickly and easily without having
to do a lot of analysis and a lot of guessing and things like that. And that's what this does.
In my mind, I think of it as a data miner
that anyone can implement and use in a small business
and compete pretty well against large
companies that have substantially more resources and better ideology,
better methodology of how they do predictions.
It's a simple little program.
I'll go through and do it. I don't consider... does anybody want to see a demo of it?
Do you need to see a Java code of it?
How about if I do it at the end? I don't want to break out of this thing right now.
But to me, it's really simple. It really is. It took me about 4 or 5 days to write it.
Anybody could write it with basic programming knowledge.
The key to it is the business concepts behind it and the application of those things.
Okay, here's an example of one of the first
graphs that I ran after I built the tool. This is the first T-Cert question.
Can anybody tell me what this pattern is? What's driving this pattern?
That's true. Okay, this is a graph made off of the keyword
of work, but can you tell me why it's so rhythmic?
Exactly. People don't talk about work on Saturday and Sunday.
That's exactly right. So, T-Cert. I don't want to take those home.
Saturday, Sunday.
Monday, Tuesday, Wednesday, Thursday, Friday. And look, Wednesday really is the hump day.
This is
a small example that there really is data that you can extract out of the
Internet. I use Deja News right now because Deja News, I've used it for several years
because it gives me a time history and I don't have to previously know about
a concept. You know, Deja News is being recorded all the
time. So, let's say RSS comes along. And I'm not paying attention. And a year later I
decide, oh, I should have been paying attention. I want to see what RSS is doing. Deja News
has a history that I can go back and I can backfill it. You can see this graph. The date here is
April 2002. I ran this graph two years ago.
Okay, I was able to backfill it and figure out what was
going on. The search engines, I have done pieces to
do the same kind of analysis against the search engines, against Google, against Yahoo,
a couple of the others, but they're very
highly skewed to the present. You can recreate, you can
make a crude approximation of the same data that you can get out of Deja News, but
it's much less granular and it's just, you've got to do manipulations, it's really goofy.
The search engines are very highly skewed to the present. You can't
backfill data without doing a lot of analysis for more than 90 days or maybe a
year. That's another key, using Deja News. Using some
kind of pre-recorded data that had no knowledge of the fact
that a trend was going on. Here's Easter Bunny,
same thing. You can see every year, along comes Easter, people talk about
Easter Bunny. The data is there waiting to be pulled out, you just have to know how
to do it. Here's Comdex. Now, Comdex
got cancelled in 2004, right? Actually, this isn't my new graph.
Comdex got cancelled right here. If I had had this graph
last year, like I did, is there any doubt in my mind that Comdex is going to get cancelled?
There's not a lot of doubt. It was a predictable event.
The fact that it got cancelled in 2005, same thing, a predictable event. You could have made
money on it if you'd gone to the bookies and said, I think Comdex is going to be cancelled.
Very clear trend, starting right here.
Each successive high is lower. Does anybody do technical analysis
of stocks? This is a very similar analogy.
Each successive high never reaches the previous high, so we're clearly in a downtrend.
Now, here's a graph. I went back later. You know, I'm not...
I don't care about money a lot, but I wish I had.
I went back and I retrofitted stuff on here, and I did keyword searches
against stock prices of the bubble. The green is the stock
price of Siebel. The blue is the frequency word count out of
Deje News at the same time period, and what I did was I matched the amplitudes.
You can see there's a... for Siebel, there's a very tight correlation. I mean, this is like a 90...
you know, for the statistical people, this is close to a 90% correlation.
Now, in this particular example, the word count does lag
the actuality, so maybe it doesn't help you so much, but that's
not always the case. Let's take a look at Palladium. Now, this is the one I really
wish I had done in 2000. You can see
clearly that there are rumors about Palladium's price spiking
almost two years before it actually happens.
Okay, right here. Although, if you really want to get technical and
drive it back, you can say that maybe this was the peak, but that's still a year. A year
before Platinum peaked, it was... you could have figured out
that there was a good chance that Platinum was going to spike. And also, if you take a look,
that when Platinum collapses, the keyword told you that it was going to collapse
before it happened almost a year ahead of time. That's money.
There's money right there. There are large companies that are doing all this analysis right now.
The fact that they're doing this analysis... okay, you know, there's guys
like me floating all around. I'm not really a
large corporation guy, but there are guys in those corporations right now that are doing this and they understand
all these concepts. And they've been piling onto the internet over the last five years
and what they've done is they've cut the margin of where you can make the money.
The latency is shrinking over the last four or five years. This is a very high latency
a year. That's amazing to me. If you look at some of the later graphs that
happened two or three years later, the latency has shrunk to like three months,
four months, things like that. Sometimes it's down to like a month.
Here's another example.
Applied materials. There was a latency. Applied materials was clearly
in an uptrend years before the price spike.
And the keyword count told you that it was collapsing
before the stock collapsed by about a quarter.
And if you look right here, you can see that now the latency has disappeared out of it. There's a latency here
in 2000 and as we go along, right about here, now there's no latency anymore.
That's an excellent question.
I've got a couple of other graphs that will show that it's very hard to determine and that's where there really is.
There is no money to be made in mechanical
skills. You can't make money
in those things, especially in a global market. You have to move into something
that requires some kind of artistic ability. And this particular
mechanism right here does require a fair amount of interpretation
and analysis that sometimes you're wrong. I mean, it's just the way it goes.
In the case of Palladium, there were rumors flowing out of Russia
that they were having problems and that they weren't going to make deliveries.
So in that case, it was the lack of metal that drove the price spike.
But you're absolutely right. There are cases where the rumor precedes
and sometimes, you know, I'm not so sure that there isn't guys at some of the companies that are actually spiking
the internet now in order to create an artificial demand in order to
offload their stock or whatever it is. So that's something you've got to be careful of.
I didn't run Enron. I didn't have any
interest in Enron. I just looked at the dot com boom. But you're more than happy
if you want the source code, you're more than happy to run Enron.
Okay, JBuilder. How many programmers do we have in here?
How many JBuilder programmers?
Ah, interesting. You guys are all assembly, low level, right?
What kind of languages do we have in here?
Well, JBuilder, that's okay.
This is a graph of JBuilder
and it basically, you can see an S curve right here. The inflection point is roughly
right here. And typically what you see, at the inflection point, your rate of change is
infinite. As you're switching over from a rising to a falling rate of change,
there's a spike there of infinite demand, theoretically. And you can see it reflected
in this graph very clearly. There's a huge spike right at the time that
JBuilder is undergoing its inflection point. And by about this time
right here, you should have been out of JBuilder and you should have been doing something else. Although
it's only been in the past year that it finally came to fruition. So there was a
lag time of about two years before JBuilder finally died off and
Borland had to admit it.
No, you have to have a keyword
that has uniqueness.
Well, you have to do qualifiers. And I've got a couple other graphs in there further on
where you'll see that, like ColdFusion. ColdFusion is a perfect one. I ran a graph on ColdFusion
because I saw an article on it and there is some evidence to support that ColdFusion is real.
So I ran a graph and I got this really strange graph that didn't make any sense to me.
And about three days later, after I published it, I found out, oh right, it's
ColdFusion the programming language. So it kind of made me look like a fool.
So what I did for that graph was I added a new qualifier, ColdFusion
Physics. And that gave me a much better picture of what was really going on.
So yeah, you've got to be careful about keyword picks, which I have a little bit later on.
Delphi. Now this is a real life example of something that I did in my own life.
I got into Delphi around 95. I did programming
on it. I wrote the wave up. And about 98 it was clear to me that this
peak was over and I needed to do something different. So right in here I started
learning Java. I went to Sun's conference. I started learning a whole
different ideology and methodology of the way I did programs. And I switched over here and I missed
the crash. There's a lot of Delphi programmers. Delphi is an extremely good language. I like
it a lot. It's the best thing I ever used. I can't make money in it anymore.
And there's a lot of ex-Delphi programmers that found that out the hard way.
This, for some reason,
I put this up and this is like the single most red graph on my site.
The guys are just fanatical about scripting languages.
It's very strange. Here you see
Perl. Very clear rise. Very clear fall.
Should have got out of it right around here. Maybe. Maybe should have.
PHP. Same thing. Rises. Inflection
points are roughly out here. It's not a true S-curve. It's not identical. But in real life that never
happens because you get all kinds of variations. You just have to kind of guesstimate between peak and fall.
And then Python and Ruby.
Python, this doesn't show enough detail, but Python to me
is the choice out of these four. If you were going to have to do a project right now
I would choose Python. Assuming I could fit it into
the requirements of the project. I mean you always get languages that fit a certain set of things.
And you can't just choose a language in order to boost
your employability. You have to look at what you're doing for the customer.
There's also a very interesting thing that happens in the past year
in that all scripting languages have suddenly got a high level of interest
out of nowhere. And I don't know what drives that.
YMAX. How many people know YMAX?
YMAX is going to be big. I believe YMAX is going to
be a big technology. YMAX basically replaces 802.11b
with the same
imagine 802.11b with a 30 mile range.
And also I think, you guys can correct me
I'm far from perfect. You can correct me if I'm wrong.
But I believe that YMAX already has permission to be able to tap into cell phone
networks. So you're going to see a huge amount of grassroots competition
against the major wireless carriers from people setting up their own shops
in YMAX. I can easily imagine someone getting a few million dollars
of capital, dumping a few YMAX things in the center of maybe 30 cities
and just cutting the hell out of Verizon's market or
Singular's market.
Blogs. This is an interesting one.
This is a little bit more
This is a question. This is a t-shirt question.
You can see pretty clearly here. This is the moving average. You got to do a moving average.
There's too many spikes that happen on the internet. Too many weird nuances.
But basically by looking at a moving average you can get an idea. And we can see that blogs are clearly rising since 2002.
And it was probably a good technology to get on around January 2004, which a lot of people did.
Did anybody read the article that was in
USA Today or one of the major papers about two months ago where they said that
27% of the public does not know what a blog is. Did anybody read that?
Now tie those two numbers together. I have an inflection graph
here, right? It hasn't reached a peak point yet. I know it's an
exponential function and I know that 27% of the people have not
been exposed to the meme yet. What kind of
thing can I derive from that? I can make a prediction from that.
T-shirt. Anybody that can tell me gets a t-shirt. Anybody that makes a good
guess gets a t-shirt.
Can you give me more detail? I don't want to go down the calculus path.
It's too much stuff for me. I took what I needed from it and I threw the rest away.
Anybody else? Down there. Black. Black shirt.
It's going to peak within the next year.
T-shirt.
It's not going to be a big money maker.
What you'll see is you'll see a lot of hysteria in the media over the next year
and then that'll be it. It'll have peaked and it'll be ramping down. That doesn't mean you can't make
money in it. All it is is an indicator. There are technologies that have peaked that you can
still make money in. That's one thing I don't want to tell you
is that if it goes through an inflection point and it peaks that you can't make money in anymore
it's a warning indicator that tells you you need to look at the surroundings
you need to look at your competition, you need to look at your skill set and make
a knowledgeable decision about what you're going to do. You can make money in post
like J2EE. I don't know, probably nobody here has read my site. I put a
controversial article out about 3 months ago
that got a lot of Java people riled up which basically says that J2EE is about to go
through its inflection point. That it's peaked out as a technology. I have no
interest in moving on to another technology right now because I still make money in it
and I'll probably ride it for 3 or 4 more years but if I were an entry level
person getting into the market I don't think I would choose J2EE because
there's already going to be a bunch of guys like me that have experience and knowledge that are going to compete against you
and the market isn't expanding anymore.
Now, it's time for the half time here.
How many people think so far that this
has been more interesting than you thought it would be? Oh, that's so
gratifying. I mean, I kind of took this
as a lark. I sent in like a basic program, you know, basic speech
People that have looked on the disk realize now that it's like about
one third of what I'm doing right now. It was like a perfunctory effort and then I decided
a couple days ago, well, I really need to do something serious on this.
So that's nice. That's what was my aim was to get, to have people interested and have some kind of
value out of this presentation.
Wi-Fi phone. How many people know Wi-Fi phone? That's another big one.
It's going to be tied into WiMAX, I think.
Okay, so we already talked about this. Inflection point tends to be at about the 50% saturation
point. That's not always true. That's just a rule of thumb. It's very easy to get
and we'll look at another graph here later on where you get a quick ramp up
but the infrastructure is not in place to support a technology. There's two technologies
that tried to go through inflection points. They tried to go through a
growth phase and the support just wasn't there. One of them is Bluetooth. I was here two years ago
and there was a lot of Bluetooth fervor. There was a lot of Bluetooth hacking, things like that.
So I went back and I did a graph on Bluetooth and I was surprised to see that Bluetooth
tried to take off three or four years ago and it died out because there wasn't
a critical mass to make it go. And then within the last two years
it's picked up a lot. So a lot of times you'll see new technologies sputter and try and
start and then they'll die out because there's some missing factor there. Service-oriented
architecture is another one. It came out around 2000, 2001, the buzzword,
all that, and it tried to take off and it died. But for the past
17 or 18 months it's been rising steadily. It's now something that
organizations either understand and feel that they need or
there's some critical component there that makes it viable.
We already did the blogging. RSS feeds.
This is very highly correlated to blogs. Same difference. By doing a search on
and you can see here I used a qualifier. RSS RDF. I don't know if
everybody can see that or not. People in back probably can't. So I used a
qualified word here because RSS wasn't a good search.
And now we're going to talk about the
problems with this. The biggest single problem that I run into is that I'm part
of the meme. How do I know to search for something? How do I know about RSS?
I read it in an article. I'm part of the meme propagation and statistically
I will learn about the meme at about the 50% mark which means I can't make any money in it.
So I've got to figure out a way to
track the meme before it gets to the 50% mark. I don't want to be like everybody
else and be the average. So that's one problem
with the methodology I'm using right now. What it requires is me to be reading articles all the time
I have to spend a lot of time trying to be at the forefront of technology
when there are easier ways to do it. It is subject to manipulation although
not anywhere near as much as you would think. When you do a search off Deja News you are doing a search
against hundreds of thousands and millions of people. It's very
hard to inject a meme and create an artificial result. I know because I've tried.
It's not statistically rigorous. There's a lot of criticism from
people that are mathematicians. I've heard a lot of criticism about it and it is kind of flaky
but it's worked well for me for 13 years. It's helped me make
a lot of decisions where I could have easily gone off onto a bad path and ended up
in another industry. Maybe that would be good, maybe that would be bad but
that's not the path I chose. It works very well for
technical trends. It does not work as well for social trends. It doesn't work
for things that have lifetimes of
20 years or 30 years. It works very well for a range of 4 to 5 years and we'll
look at why in a few minutes. If I were
serious about using this as a technology to drive my company
and make decisions with, I would throw away the current methodology of using Deja News.
I would start capturing RSS feeds and I would build my own
database and that will allow me to capture trends as
they happen in real time and I guarantee you that there's people out there right now that are doing it.
I did it. I wrote a little piece about 6 months ago.
It has some technical problems but you can easily get around them if you spend
enough time on it. I have a lot of other things to do but there is a lot of value
to be done in applying this methodology to pulling down real time
RSS feeds, building a database and looking at rates of change that are
happening in keywords. I believe there's a lot of value there.
We already talked about skewing factor. One of the problems is
you lose abstract relationships with the way you store the RSS.
There's some things there but they can be gotten around.
Okay, so now we did the business process.
We did a tactical implementation that gives us real decision making tools.
Now what we're going to do is we're going to push this back up into a strategic framework
that will let us make decisions about more than just about what technology
I choose. And this, I admit, this is going to be kind of flaky.
One of the big justifications for Moore's Law
people like to quote Moore's Law. Everybody knows Moore's Law, right?
How many people don't know Moore's Law?
It's a trade off. I can't let Moore's Law go.
It gets quoted a lot because it's used as a justification
that the number of transistors will double every year
so there will always be an IT industry that will continue booming.
And I would agree with that
if there were no other factors.
People have a finite amount of time and effort to spend on anything.
Geeks, techno geeks,
they like technical things, they like trends, they like
detail, but ultimately people are social animals.
They will spend time on things that they enjoy
or things that they're forced to do. Other things they will throw
away in favor of things that they enjoy or things that they're forced to do.
And this is part of the reason why the Memminer works off of Deja News. People talk about
things that they enjoy or that they're forced to do.
For the average person, they spend their time sleeping,
working, I'm glad I got a laugh, that's good, shopping,
eating, having sex. The range of that
variance is not a lot from person to person on average.
So if I want somebody to read the internet two hours a day and I get
a certain level of growth, now to double that, I've got to have them read the internet
four hours a day, it's not going to happen. They've got other things they want to go do.
They want to go eat, they want to go have sex, they want to go visit Las Vegas.
So there is a fundamental limiting factor in the growth of the internet based on what people's time is.
There's other factors on top of that. Now for some people,
you do get a wide variance.
For instance...
But I don't think we're going to see a society that looks like this.
Now we're going to look at Deja News again, and what we're going to do
is we're going to invert. What I was looking for was uniqueness, right?
What we're going to look for now is the ether of
Deja News. We're going to look for the rate of growth of Deja News
abstracted of all uniqueness.
What I'm going to do is I'm going to use keywords that have no
identifying ability, that are used in anything, anytime.
This, have, one, all, them, will, other. That comes out of the top number
of people that do crypto analysis. E-T-A-O-N-R-I-S-H-D, right?
That's what you do a frequency analysis to be able to crack codes that are not well done.
Well, you can do the same thing with keywords. What I did was I went through the top 100 keywords
and I picked out ones that were fairly generic, and then I did a search
and you can see very clearly exponential growth in Deja News.
This is one reason why the MemMiner works is because over any small period
like from here to here, it looks linear.
There's not skew being introduced by Deja News into the uniqueness of a meme.
There's a little bit, but not enough to matter.
Okay, so this is what the ether of Deja News looks like.
The transport.
If you do a really detailed search, you can pick out events that cause excitement on the internet.
And these are admittedly, I guessed at them, I think I'm right on most of them.
JFK crash, a lot of people can't see this, but there's a spike right here.
Y2K causes a spike, there's a spike here I didn't hunt down.
Bush election 2000, that definitely caused a controversy.
9-11 causes a big spike. What's interesting about 9-11
is that if you do keyword searches
on other terms, you get spikes.
If you look for death around that time period, you get a spike. If you look for
peace, you get a spike. What you don't get a spike in...
Okay, remember, we're talking about people that have finite bandwidth, right?
They're using more bandwidth to talk about 9-1-1, they're using more bandwidth to talk about death,
they're using more bandwidth to talk about peace. What is being cut out
of the equation to free up that bandwidth for those things?
Can anybody give me an idea or a guess? It took me weeks to figure this out.
No, there's, well, there's a particular thing that
it cuts into, and I think it'll make sense to everybody once they hear.
No. Work, who said work?
T-shirt. Work goes down. And I remember 9-1-1, what did I do on 9-1-1?
We all sat around talking, we didn't do any work.
Now, here's the bad news.
Deja News gets very flaky after 2004.
I got a couple theories about it.
I have three theories about it.
One is the shift of blogging, clearly. What people were using Deja News for, they now use blogging for.
But that does not account for the rate of change that you see here. This is almost instantaneous.
You see the moving average here, but the actual numbers,
it changes within like two months, and I know that's not happening. Blogging is taking away chunks.
Google changes their methodology, their search methodology.
If I go back and I run a query two or three months later, it changes. They're doing subtle
shifts in the way they do things, and it shows up. And I do think that there's some kind of
error in the way that they're manipulating or accounting for things
after 2004. But I think a lot of what you're seeing
is the inflection point in the internet itself.
If you read enough articles, you see a couple articles came up in the past year or two
that talk about a slowing rate of growth in the internet. I think that that's what you're seeing.
I think you're seeing a slowing rate of growth in the internet, and it will impact
all the other industries that are associated with that, mostly consumer-related industries.
Chip industry, business software, it won't affect
niche markets so much. And it probably won't affect security very much.
And now, we're going to take a look at a couple concepts
to make strategic
decisions about how you're going to spend your resources.
Very abstract concept. What caused the IT boom?
If I were to create a market basket of all the items
that create a cost of how I get information, chips cost
me money, people cost me money, I have to develop methodologies, there's a lot of
communications costs, I could create a market basket of the cost of
information, assign it to a unit of information, and I would be able to get a rough number
that each unit of information cost me a certain amount of money.
Now, up until 2000, everybody was making money, right? That's because
I have a price of information. I'm selling that information to somebody somewhere
and I'm making a profit margin on it. As we get down to 2000,
cost of information is not decreasing like it used to,
and companies that are used to this curve, you know, they get into this mindset of seeing this,
they overran it and they went into debt and they went out of business.
Some companies veered off to the side and adapted and survived.
But ultimately, if the cost of information stops falling,
there are fundamental changes in the marketplace that companies are going to have to adapt to.
And ultimately what you have to do, because market forces
force you to adapt to a pricing model that the public sets, or governments,
you have to find some way to drop your cost of information.
And that's why you see outsourcing over the last 3 or 4 years. Companies have to.
They have no choice. Otherwise they'll go out of business, because there is a ceiling on what they can charge.
That's a very strategic concept.
It sets the stage. You can't do anything with it directly. It's not a tool.
And this is another picture that shows you roughly the same thing
from an economic standpoint. As a market matures, your marginal cost
of processing stuff rises. Your marginal return falls.
You eventually reach a point where there is no profit margin, as you pick up all the easy things.
What that means is that there is money in hard things.
And the same difference. Now this is a theory of my own, just from watching the market.
Watching the IT market for the last 10 or 15 years,
the market rewarded diversity. As I created programs
that did more and more stuff like Word, it does all the bells and whistles and stuff,
I got a bigger and bigger share of the market, and people bought my product.
But once I reached the point, the inflection point,
the inflection point punishes diversity, and it rewards cost-cutting.
And that's the point that we're at right now, for consumer-related industries.
Like I said, there are niche markets that don't get impacted this way right now.
So, from my standpoint, this is what I see.
I see the up-cycle favoring diversity, and everyone pursuing that,
and they get into a mindset over the last 20 years that diversity, diversity, diversity,
I want to create a diverse product and get as much market share as possible.
We've seen a fundamental change around 2000, which now punishes diversity and rewards cost.
And most people's minds are still in this state.
I don't know if I should go into this or not.
This is the single most read picture on my site.
And it surprised me a little bit.
In a nutshell,
diversity functions on contexts that are mismatched.
If I had two clones, and they had identical information sets,
and they know the exact same information,
and they've had the exact same experiences, their communications and teamwork will be perfect.
There will be no miscommunications, correct?
As those clones have different life experiences,
their contexts will shift apart, and there will be a set in the middle that they share,
but there will be unique features that they each have, and then you'll start getting miscommunications.
If I wanted to cut my cost of information
in teamwork or software development,
what I would do is I would try to get contexts to be
as matched as possible within a team framework.
Like I said, this is very high level, and there's some other things I played around with it,
but I put it in there just to maybe get people thinking about ways that they could
drive their teams or their software development towards a
competitive methodology, and what you have to focus on
is reducing transaction costs internally in your organization.
There's a couple other things here. I'm sitting in a microbrewery, and I realized, wow, microbreweries are a lot like the software industry
in certain ways. The company that I work for right now
basically resells IBM software, but what we do
is we tweak it to a very minimum level, five or ten percent,
and match it to that set of customers that are willing to deal with
changing their business processes to map to what the software does, and that's a very
efficient model that I think can last over time. I don't think a lot of very
customized software, it's going to have a very hard time surviving in business software over the next
ten years. And we'll
let that go. I'll go ahead and I'll take some questions.
The question is, do I think that the difference between
the dichotomy between diversity and cost-cutting is driving some of the open source movement?
I don't know. I don't... I've looked at the open source movement
and I still honestly don't have an opinion about it one way or the other. I can't get a reading
on it out of the graphs that I've run. I can't speak to it one way or the other.
I haven't done any analysis of that, but I would say companies that are doing Six Sigma are probably
unless they're in like a financial industry, I think they're probably
at a competitive disadvantage. I think they're producing quality that the market isn't going to reward.
That's a guess off the top of my head.
Yeah, you can use the search engines. You can
get a lot of the same data out of Google. It comes out in a different format
and you've got to massage it and you've got to work with it and it's very crude. It doesn't give you the same
detail. But yeah, you can pull it out
of other sources. I used to use Monster. I used to run queries against Monster. I used to use Dice.
I used to use Miscellaneous Jobs News Group for people that remember that far back.
No, I haven't done anything on security
because that's not my field. One more.
A lot of what you just saw, not in so much detail.
There's more business here than I have on my website because I haven't updated for a while.
There's a lot more graphs that show you more detail
in different industries like religion. I did some queries on religion,
social trends, things like that. So you can get an idea of different things. And actually, I did mean to bring that up.
I'm not out of time yet, but
we can look at a couple of those real quick.
Or maybe not. There's no doubt in my mind
after watching what's going on in the marketplace, there are a lot of people
that are smarter than me. MBA program teaches all these things. I haven't been through MBA.
I did a little bit of an MBA, but they teach you these kind of concepts and there are guys out there that are applying them.
I have no doubts. You can see it in the way the information has been changing over the last five years.
There's someone that's taken a chunk, taken a slice of the profit margin out of it.
Let's see, what can we look at?
Oh, here's saving J2E. If you go through here, you can see that J2E
looks a lot like it's topping out.
Yeah, these are good right here, Republicans.
Does anybody care about the federal deficit?
You bet people care. Whether the media tells you they care or not, it's being talked about a lot.
Trade deficit. Does anybody care about the trade deficit?
Media doesn't talk about it, but people are talking about it.
You can get a lot of interesting information that runs
counter to what the media is telling you. As a closing comment, I'll tell you
that when you see Alan Greenspan talking on TV, don't listen to his words
anymore. Watch the beads of perspiration on his forehead.
That's what matters. And that's it.
I got two more t-shirts up here. Two more t-shirts if people want to nab them.
Oh, am I out of time?
Do I have like two more minutes?
Real meme. R-E-A-L-M-E-M-E.
Yeah, yeah, that's what I'm doing right now. I'm going to run it.
I got like a minute.
It's a very simple program. Send me an email and I'll send you the base to it.
I'm not going to support it because it's got a couple bugs in it that I work around because I don't want to spend time on it.
But we'll run one real quick here that might be funny.
No, I think there's profit.
I think ultimately what you're going to be driven into is, at least in the United States,
I think you're going to be driven into niche markets.
In a single phrase.
Niche markets that are supported by small industries that have high profit margins
that are willing to pay for some level of customized software development.
But that's a trend that's going to run over, I think, five to ten years.
It's hard to say. I do think, for anybody that cares,
I do think J2EE has peaked. And if I were entering the market, I don't think I would put any effort
into trying to learn J2EE as an enterprise system.
I think there's alternatives that are on the horizon.
I think your time would be better spent in other things that are built upon the infrastructure
of the existing S-curve that's already run.
Oh, I was going to mention Newbridge.
The Newberry guys in there. The Newberry guys in there that are running that
that tracker, that wireless tracker, that's an example of a technology that's built on
a foundation that's been built over the last ten years. That's the kind of thing I would look at.
Okay.
I look for it. I see it. It comes up in different places
in different ways. It's intuitive. It's from seeing it so often.
It happens around 2004, 2005.
Gosh, I'm really sorry here. One more minute.
Well, if I have a good net connection
and I'm running it threaded, probably 20, 30 seconds.
Yeah, I have done run RFID and that's on my site.
It's in an uptrend. It's in a pre-inflection state.
Definitely worth putting money into.
Hmm?
That's a whole other session. That's a political session.
No, I don't blame you.
Anyway, here's Paris Hilton.
She's still got some life there. I'm surprised. I didn't expect it.
Okay, thank you.