Trevor Oakley

Trevor Oakley Sites

News

I am attending BigFish run by Doug Richards (ex-Dragon Den investor).

Web Business Models

The Internet is basically a network of computers and also a network of people. There is little ground breaking software, eg in artificial intelligence or 3D control. Therefore, Web business models do at this time focus on bringing together people rather than being at end delivery point for a service. For example, when using a search box, a consumer is trying to get something else and the search is a means to getting that. The same logic applies to any of the business models featured below.

A key in the present models has been to bring together two parties via a Web service and not the service itself. the two parties can both be consumers, for example with Facebook or Youtube, or just one is a consumer, eg with a search engine.

Companies trying to deliver a wholly based end-service have generally failed or struggled more than others. Amazon struggled for years trying to sell alone, and today Amazon sells extensively via partners. That is vindication that a business model based on a website acting alone will not work.

Early shopping based models relied on Click and Mortar under which a store would sell online. But such ventures have met with limited success. There are no known reasons why large stores have largerly failed to penetrate online markets.

Business Models Not Realised

The web still offers a range of chances to make success. Education is largley not realised online. News content is still very fragmented and forums, blogs still carry a lot of weight for news. No company has dominated that market.

My personal interest is VRM which is a system bringing together buyers and sellers. This concept extends the original ebay concept to all products and also introduces a personal service.

How business models were realised

The common factor has been raising large amounts of capital early on. This has been combined with a early management team of postgraduates or at least university focused people.

After money was raised, the media outlets were coaxed into writing volumes about the start up after which the name would be branded.

My Personal Interests

To provide an education based service online, 3D Graphics, Artificial Intelligence, and a VRM based service for shopping.

Education would require an interactive service for information exchange

The 3D and AI aspects would relate to developing software which could be licensed to various companies to measure website performance, eg consumer behaviour. Such software would measure behavioural patterns and how they relate to buying services or products. Finally, a business goal of any business is to sell something. So buying patterns matter.

Dramatic Failures

Webvan, Pets.com, Infoseek, Lycos (once challenging Yahoo!) now employs under a 100 people.

Key Technologies for the Web

I have found the following very useful:

I have found the IDE framework for .NET very useful. The db4o database system is very good for storing objects.

More on VRM and related issues

For all your shopping see Merrows. merrows

Intelligent Shopping

I started work on the new intelligent shopping system. My first task is to write the bot, and I found an excellent link for this: Heaton Research. I have the book also, and the book is great. I am using C# as it is an easy language. Proposed work would involve a VRM project.

Introduction

I started thinking about a concept of "democratic shopping" which extends ideas already used by digg as a democratic news site, and google (democratic rating of sites via links). The shopping issue is complex for the following reasons:
  1. Products are rebranded (same product renamed to avoid price matching).
  2. Brands hide the true identity of a product (eg something made in China for a EU company is then sold under the EU company name).
  3. Products have a level of exposure (image) which defines their worth in the market. This is related to advertising, brand awareness.
  4. Corporate deals, eg price fixing, control prices and reduce choice for the customer. The customer simply thinks the price is good due to clever price awareness ("we are the cheapest").
  5. Controlled distribution which cuts out competition and reduces choice, combined with the above means customers pay more and get less choice

I started as a working site sayworth which really is a forerunner of a democratic shopping website allowing customers to vote on products with their feet, ie by buying them. I intend to use Semantic Web technologies as follows:

  1. Relating user interest to a product, eg buy, reject, return, ignore, laugh at, recommend, refer
  2. Relating products themselves by user interest. For example, customer buys A and then B,C so A is related to B and C since there is a purchase relationship. If customer buys A and then rejects (ie looks at the products but does not buy) B,C the relationship is a reject relationship.
  3. Cutting through the brand to show the real product. This is done by effectively relating the same products that have different brand names.
  4. Examining real prices, prices are often hiding by 50% deals (after a 100% increase).
  5. Relating products to an actual customer profile. For example, an eskimo buying a coat has a different requirement to a Californian buying a coat. This is probably one of the most interesting problems to solve. One solution is simpy by traffic analysis and shaping buying patterns. For example, a person buying very heavy thermo boats may also want similar clothes.
  6. Recommendations based on profiles, product relationships, and product successes. So a customer buys A, and then B,C are recommended.

I've been checking into some etymology sites (word origins) and I found a great one for weird words. There is also an online dictionary for etymology.

I have been considering a word for my new venture and I decided on yafful as a working name. It means armful of hay.

Notes on Software

I have started coded some basic c# bots and discovered the need for an OODBMS very quickly. So I am considering switching to Java to getting OO interfaces.

Some OODBMS links show what is available.

VRM - Vendor Relationship Management

I was watching Intruders TV and saw an interesting interview about VRM from a VC manager. VRM is about relating vendors to customers. This seems to describe what I am trying to do in my project. Basically the issue is one of identification of user interests. See the RWW entry about VRM.

Making a business

How to make a business from VRM? There is a good Youtube video on VRM. The basic concept is to build a shopping cart for users. The user then selects what to buy. The cart is then made available to vendors (eg retailers) and they put in a price for the purchase request. Therefore, they bid for the business.

This is different to a price comparison website in that the customer selects what to buy as a group buy, eg several items and then a vendors proposes a price and delivery time. Price comparison sites work well for one product on a link, but they fail when trying to sell unbranded items, and there is no link to the user (eg what do I want). Clearly this is all Web 3 related and this will finally proceed to the systems automatically buying what you need based on your wants.

Notes on SW

One key aspect of the SW is relating references by time and also events.I saw an article which talked about someone liking ice cream but not all the time. I am a frequent buyer at Amazon, and I notice that Amazon will bring up recommendations based on my searches from weeks or even months ago. One key rule of buying, is that it is a now business. So recommendations need to refer to now.

Regarding the ice cream point, it relates to the time of day (eg noon), the weather, someone's mood, someone's location, the social occasion, and other factors. You can model this in data, but it is impossible to use as too many factors exist and processing times would be too long. All you can do is build a self-adpative system that just looks at when people purchased ice cream and matches their pattern of behaviour with someone who did not. So the system takes only decisions already made and projects them for a new instance of a decision. It will not pre-judge based on some kind of hard coded logic, eg if hot buy ice cream. But if sureBuy then recommend buy.

Using these triples we can enrich data and add semantics to it. Now bring in the very human factor of mood. I love ice cream. Does that mean I love it all the time? No it doesn’t. Actually, I rarely eat ice cream, but I do when I feel like it. How can this be modeled into data? Depending on whether or not I had a great cup of coffee in the morning I might feel differently about ice cream in the afternoon etc. etc.

Interesting websites

A good site is Intruders which has video interviews of CEO startups. I saw veedow which is a different take on a product recommendation website. Veedow uses user recommendations to rate a product. I saw a similar idea expressed at storycode in which users classified books. The site was featured on Dragons Den.

Such business models like all these type of models can only work with a high user recommendation activity. Systems depending on user comments require a large database of comments so an effective average can be found and the result useful. Otherwise an odd few comments, for example faked by competitors or even recommendating your own products, will skew the results. I estimate you need 100 responses for a product to have a meaningful average.

Do you want to help?

I am eager to hear from anyone skilled in high technologies to help with the above. Key skills
  1. Applications: .Net, C#, RDF, OWL, XML,HTML, SQL, LINQ, ORACLE, DB2, SQL SERVER, UNIX, Windows
  2. Systems: Load Bearing technologies, handling large databases, high volumes
  3. Marketing: Knowledge of sales, branding, SEO, Promotions
  4. Expert Knowledge: Knowledge of retail, wholesale, import/export, brand identity (for products), sales (for products), logisitics
  5. Linguistics/Intelligence: Knowledge of natural language based systems, artificial intelligence, control systems (eg self adapative systems, self tuning, fuzzy logic, solving complex equations)

Semantics and Search

There is a good semantic search engine at Falcon. It has a lot of OWL and RDF entries.

Semantic Web

Radar Networks - interesting link I saw from theNextWeb.Org

I was fishing around and saw the next web site, which had a good video from Radar Networks, so I am listing this site

Semantic Web Interview (audio). This is general talk.

RDF (general comments) - Long and verbose.

RDF and C# (classes for .NET). This is not well documented.

Intruders TV. This is an excellent site for seeing video interviews with CEO's of start ups.

Meetup.com A good site for arranging meetings for entrepreneurs. There are several groups for the Web and C#.

Berners Lee writing in 1998 about the Semantic Web. This is a basic document.

Wikipedia and the Semantic Web. This is a standard documnent.

Blog Search from Google. This is an interesting first place to look for Semantic Web comments.

A semantic web report. You have to buy the report.

RDF Triple Language. This is dated November 2007. This is an abstract document.

SW Blog - Good

Webpedia for SW

SW and publishing

Links for LINQ Videos A great intro to LINQ and C# and how the var object is now used in C# (3.0). This object allows queries to be done without stating the type and hence LINQ/C# determines the type.

People skilled in logic were historically called logicians, or philosophers. The fields of mathematics, control theory/systems, philosophy, software engineering, artificial intelligence have several commonalities and overlapping areas.

People have been trying to understand the human brain for a long time in the hope of artificially reproducing its functions. Fields in Neural Computing tackle this problem, and also the fields involved with self-tuning systems (algorithms which write themselves).

Pure logic presents the principles by which a conclusion can be reached, eg by deduction.

Modern computers mostly process data very quickly (much more quickly than the human brain), and this gives the appearance of human like processes. In fact, computers are only quick - they are not intelligent.

Aristotle
Boole
Leibniz
The Internet is probably the most recent major commercialisation of computer technology. Today, XML, ORACLE, FLASH (from Macromedia), PHP, IPIX, and 3D are the most important technologies in play. There is a technical page explaining more.

Some Interesting Sites