IMAGE LABELER

Google's Image Labeler... As I mentioned here, there is still a lot of work needed to support relevancy for search. I have no idea how long Google's Image Labeler has been around, but I think this tool will help clarify my crazy little illustration from a few posts back



So, maybe there is a chance Macro Marketing will succeed. ;)

Below is a screen grab of my partners scores. We labeled things fairly well compared to other guests.



I can't believe my partner passed on an image of Rage Against the Machine!

Google / Yahoo

My concern lies with the [potential] homogenization of creative and ad *spaces* as we would see even less flexibility within search, to create unique experiences for users. I've expressed a few thoughts before regarding the internet as an operating system versus a neutral network for open source development.

I think it could potentially force advertisers to develop tighter ad networks with portal sites and aggregated content/industry specific sites/blogs. It could cause a predominate *first tiered* network for the internet built exclusively on ad spaces. Which in-turn, could become a wasteland or repository of useless content. I realize this might sound like a faux-future problem, but its a concern.

In expressing the concern, I'll counter this argument with a potential positive;
Distribution and serving up ads might become easier for small businesses and ultimately leverage the field for the entrepreneurs and start-ups.

CREATION OF VALUE–VALUE IN CREATION

OUT-THINK EVERYONE OR OUT-CREATE THEM?

Creativity is an intellectual labor. The truth of the matter, no idea can be considered a potential failure until it's respective execution has been experienced or utilized in some fashion.

Show me something that is not art. Show me something that was not created, for a purpose. Even if it's purpose was defined after it's inception.

RAPID INNOVATION AND CREATION BASED ON EMPIRICAL OBSERVATIONS

I used this and it made me think this.
So I did this and this happened.
What will everyone think about that?
Do you like it?
Why not?
What can I do to make it more valuable to you?

Practice your creativity, or you'll never know what might work in the future.

MARKETING ACROSS MACRO REGIONS, HMMM.

Even with a successful implementation of a semantically-driven web, this will be very difficult to introduce.

Without user's contributing concise and accurate data of their own, sites will always sell advertising to the masses or be forced ( as always ) to generate better ideas with each campaign. Micro-market or not, it's still about insight and creative application of this insight. And the ability to provide relative content. Fair-enough. The semantic web claims to offer this...

Please humor this hypothetical model;

Sites would have to per cap the amount of ads served up, based on length ( :1s, :10s, :15s, :30s or 'x' amount of time–whatever ) and then prove that 'i' amount of users would see these ads. Never mind that we haven't established or proven that a user(s) 'i' will click or interact with the ad beyond an initial eye-ball acknowledgment. It's possible but, only within an existing social network. The web is to diverse and complicated to accurately refine this data.

At this point, a site could technically charge fees based on the number of users expected to see it–fair enough, quantify the users and qualify the views. But, there isn't enough physical time in the day to qualify and quantify these stats. Simply because the internet(s) are not globally/universally regulated. Consistent rules and regulations do not exist for the internet against a continuous and common revenue stream, let alone a physical timeframe. There is no global standard for money, time or products either.


What is being suggested is:

1) Pavement Shoe is gauranteed 300 views and billed accordingly, per view.

2) Nike gets 30,000 views and is only charged for those specifically.

Neither company pays for wasted eyeballs; i.e., 200 cars drive by an outdoor board and only 3% respond.–yet, we still pay the same amount of money for the board, regardless of our client's ROI.

None of this works unless users have provided accurate data within their profiles in order to serve ads up correctly. If this data is being mined based on profiles, then we're all screwed. A profile may state, "I like the sun and the moon." This data must be mined rather quickly and verified in some way as an actual ( legitimate–a real person ) profile with meta data being complied instantly, to determine relevancy. And context of the words must be concisely defined. The site would then have to compile data that demonstrates there is a market for Pavement Shoe's type of business. This can only happen if the user has been honest and true to their profile data. Let's not forget about the power and ambiguity of personal context.

Ultimately, the sites would then have to assign a monetary value to the user's respective profile and their friends; connections; network–whatever the site calls them, as well. A user would be classified by a system designed to gauge and evaluate the value of this user, based on geography and key words within their profile. Again, we're limited to the context of these words, sentences, phrases, etc. Tracking is crucial here; where is the IP coming from? Is it static or dynamic? Is it even real? Is it assigned specifically to one person? Chances are, no–of course not. ( all-though mobile technologies will change this. But that is a different question entirely )

This is a failed model considering most participant networks are still used for escapism and entertainment purposes. Granted, there are a lot of sites popping up that require validation and verifiable credentials but, who the heck wants all that info cached? The mass majority doesn't. The prosumer, doesn't–for sure. The only other option for accurate tracking is; financial data. And as of now, legally those companies cannot disclose that information for any reason.

Paid-words strive to solve this in some way but it does not give the ads ( whatever they might be called ) definitive contextual relevancy. The context of the words themselves, remains subjective. Nothing is truly obvious, especially to a program ( da bots ) that works linearly, as opposed to, episodically like our brains. *Obvious*, is always understated. We would also need to develop a database of visuals that every user in the world, has agreed represents ( roughly ) the visual context of the item.–We must all agree what color value best represents Blue and then agree what color value/shape-object represents Blue Chair.

In the end, if you're not on a community-social-participant-driven site with paid placement, a brand must create unique content and work it's way to the top of natural search. This is empirically-driven search, as the web was designed initially. The brand must develop a rich reputation. This is the reason we have so much rapid-protyping of social networks. Advertisers need street corners for prosumers to gather around. Insight, occurs here–sure, but it's being compiled through empirical observation. Data mining or ethnographical data can be researched more concisely within a controlled environment–no problem, makes sense, but this is still subject to personal context. And whether or not the user has provided accurate data about themselves initially.

The web is an aggressive medium. I must search in order to find. If we try to constrain users to a controlled environment, with an exact set of question fields, with a set and discriminant list of answers–then perhaps this model may work. However, at this point, privacy will become the new commodity versus any other commodity.



See a previous post regarding a change in morality.

"A user couldn't be involved unless they have contributed in some fashion. If this transparency is merit-based, it would neutralize the delusion of surveillance."

And to further the reasoning behind this and advance the idea of transparency, unless a user can contribute to the overall development of a brand, the merit-system fails as well. If a user could, literally, assign a picture with every word they've typed within their profile, it might be possible to then aggregate this information accurately. But this still requires a large amount of user control and ( I can't even guess as to, that amount of server power, bandwidth and hardware it would take ) to allow a user to edit this information on the fly. And then have the old data refresh instantly! Hmmm.