ISORG: Dr Antonio Cordella's Review

(Abridged) Sessions 1+2+3 (31 Mar + 5 Apr 2011) (COMPLETE)

This is a summary of the notes I took during the sessions, consolidated into 1 set of notes.  Remember, these are the sessions as seen through my "lens" and subject to my own interpretation of what I know.  If you want the real thing, make sure you attend the sessions yourself.

Overview of the course

The course looks at technology from a very precise angle. From the organisation viewpoint. About the use of technology in large organisations. Adoption and design of information systems. Organisations have specific goals. It doesn't matter what the goals are. What is important for ISORG is why and how organisations choose to employ technology to support their goals.

How do we choose technologies to support the organisation?

Because different organisations have different needs, we need to use different models to analyse these needs so that we can use the right technologies to support these needs. These models reflect different approaches to the design of information systems. The reason why we have different models is because organisations face different problems. Each problem may require a different approach.  What big organisations need is to find the right technology to suit their organisations.

Example. What would you use to write an essay? Excel or Word?

Excel can be used to write essays, but it is not the best technology to use.

The models allow us to choose the right technology to employ for the right problem for organisations. Technology not only must be designed to fit the technical needs of the organisation. It must also be used by users. A word processor still requires humans to use them. The outcome of using the systems is to address the specific problems of the organisation.

What is the best way to design IS?

First define what the problems are. Then design the technology which suits the needs of the organisation. Hence the best way is the "prototyping" approach. Create a simple system and let users use it to get feedback. Then customize the system based on this feedback. Idea of the prototype is to design the technical needs of the organisations, while the users' feedback will make the system more user-friendly. Technology is co-designed with the users. This technology will have very little implementation problems. [WT: Antonio is trying to explain how prototyping can lead to the design and implementation of socio-technical systems]

However, prototyping cannot be done for very large systems which is an assembly of different information systems.  These systems have so many inter-dependencies with other systems that they cannot be customised.  They have to be designed and then implemented into the organisation.  When implemented, they cannot be changed because they support so many procedures in the system that any change will have impact on a number of other systems. Impact analysis. Eg. air traffic control system. Can it be changed after it is implemented and operational?  No, because it is mission critical and affects too many things.

So enterprise systems cannot be "customised" once it has been designed. Hence the "optimum" approach to design (which is prototyping) cannot be used.  So we have to use another approach.

Formal IS development

This approach starts from the analysis of the organisation.  Why do we need an information system?  Maybe we don't need one.  By analysing how the organisation works, we can create a model of the organisation from which we can design an information system to support/improve the way in which the organisation used to work.

ORGANISATION - - - - - organisation analysis - - - - ->
MODEL OF ORGANISATION - - - - - information systems design - - - - ->
CONCEPTUAL DESIGN - - - - - implementation - - - - ->
ORGANISATION

[WT: You must recognise the 3 stages involved:

• Analysis (where we choose a model to analyse the organisation & its problems)
• Design (where we translate the requirements derived from the analysis into a conceptual information systems design)
• Implementation (where we develop and implement the system based on the conceptual design)

These 3 stages are inter-related.  The analytical model used during the analysis stage will affect the design of the information system, which will then affect how the systems will be implemented.  Less time spent on analysis and design generally leads to more time spent resolving problems during implementation.  More time spent on considering more parameters during analysis and design will help to reduce amount of effort spent on implementation later on.]

Each model is a different pair of glasses which see different things in the organisation.  Information systems cannot be designed on the basis of the organisation and needs directly. This will be too much to analyse.  The models simplify the the organisation and its complexities, to address only the elements they focus on.

• DM => Look at the way data is prepared and exchanged (business processes)
• DMM => Only look at the decision making processes (decision making processes)
• TCM => Only look at transaction costs (transaction processes)

There are many other models. These 3 are discussed only because they are the "most commonly" used.

Information systems are designed as a result of the analytical model chosen. When information systems are implemented, we face implementation problems. This is not surprising, since the IS is designed based on the analytical model chosen, which does not address ALL issues and aspects of the organisation. So implementation problems occurs because every system is designed only to address "very few things" of the organisation.

That is why we refer to the models as different ways to inform the design of information systems.

Example:

Problem: Optimize traffic going from a specific location to another location in London.

Analysis: Google map of London. Use the map to redesign the traffic of London. The map represents the "model of the organisation".

Design: Roads are redesigned to go straight from the starting points to the destination points in question.  Number of carriage ways are increased. This is the "design".

Implement: Is it easy to implement the solution? No, because there are lots of obstacles and interdependencies that cannot be moved. (eg. Tower of London is right on the straight line path.  Cannot be moved)

So change the representation by adding more features to the Google map. With the additional information, the analysis becomes more complex. However, if we do consider these complexities, it will reduce the amount of implementation problems.

So the 3 models are different representation of the organisation.

DM is based on the data flows. It is very "simple". Makes it very easy to design the information system. But it has more challenges during implementation. Eg. just increase carriageways. Straight forward solution. Very challenging to implement.

DMM is more complex because more things have to be taken into consideration. But we reduce the implementation problems. DMM is richer.

TCM is even more complex and richer in terms of details. Therefore we provide a bigger picture of the organisation. But the implementation problems will be much reduced.

Example. If you buy clothes from Zara, they may not fit completely, because they only have very few models. Mass production. If you tailor-make your clothes, the clothes will fit exactly. Zara is assuming all the people in the world has only a few sizes. So more implementation problems. Tailor-made clothes have less implementation problems.

A very simple representation of the information system leads to the design of very reliable information systems. The price paid is in the implementation stage. A complex representation is much harder to design because we have to take care of a lot more details, but it reduces problems in the implementation stage. The more we invest in the analysis and design, the less we have to invest in the implementation.

When IS fail, we usually blame either poor requirements analysis or poor coding. However, perhaps another answer is because we chose the wrong model to represent the organisation.

It is not possible to design information systems directly based on the organisation's needs because it is too complex and has conflicting needs. Information systems are designed for very specific things. They can only do what they are supposed to do.

Example. A Motorola Startac designed 13 years ago is not designed for sending SMS.  Nokia segments their market very well.  Their phones directly caters to the specific needs of users, eg. smartphones for executives, multimedia phones for consumers, etc.  VCRs have dual "simplified" and "advanced" menus to serve different people.

When we look at technology, we must ask ourselves, "what are these systems for?"  So when we look at enterprise systems, we must ask why they are they for

The DM, DMM, and TCM represent the intentions or the "logic" of the systems.  If we are able to understand what is the intent of the information system, we are able to understand what is behind the systems.  If we understand the logic embedded in the systems, we can understand the implementation problems and the challenges of the systems.

Technology

We must understand technology because we can use it.

Example. The iPhone and iPad represent the latest and the greatest changes to technology. The keyboard and the mouse "disappeared".

Technology is the combination of hardware and software. Information systems is a combination of these 2 components.

Hardware

Basic components: CPU, Memory, I/O

"Motherboard" is the highway. CPU is the "brain" of the computer. When the computer is turned on, the CPU bootstraps off the BIOS which tells the CPU where to look for components. The CPU reads the set of instructions stored in ROM. The ROM loads the O/S in the HDD or DVD or USB drive, etc. Once loaded, the O/S takes over the entire computer.

Operation System

The O/S defines what can be done to the computer. It allows us to explore what is happening in the machine, but that's it.  It is a low-level software that allows us to use the peripherals. Peripherals have to come with drivers.  Plug and play => plug in the new devices and the machines automatically recognises what to do with the new device. The O/S like Win 7 already contains all the drivers for most of the peripherals you may plug into the system, and so enables plug and play.

H/W features can only be used if it is recognised by the O/S. If it is not recognised, the features cannot be used. Eg. different DVDs can only be accessed in certain countries => DVD Region. It can be cracked (eg. to force the O/S to do something it is not intended to do).  Printer manufacturers sell printers to different markets. More expensive to build different hardware for different people.  So they build the same hardware, but configure the firmware to limit features and capabilities to cater to what people are willing to pay for.

The hardware is the basic fundamental. How it can be used depends on software. Windows and MacOS can run on the same hardware, but the way the machine is used is different depending on the O/S.

Layering of standards (S/W and hardware stack)
Application S/W
O/S
H/W

Microsoft argued that Internet Explorer is part of the operating system. Authorities correctly insisted that IE is an application software to prevent Microsoft from trying to monopolise the Internet.

These days, what differentiates between computer manufacturers is not the O/S or the hardware, but rather the applications that run on top of the O/S and hardware. Users prefer the iPhone because it has more applications and access to iTunes (music, video).  Nokia is failing because they did not understand the importance of the application ecosystem that sits on O/S and H/W.

Why is Skype so significant today? The technology is not new (CU-SeeMe, instant messaging, ICQ). How did Skype survive and the rest disappeared? Reason is the IP protocol.  CUCME requires fixed IP addresses. Skype separate identity of user from the IP address, which is limited to only 4billion addresses and will run out soon.  Skype is based on P2P protocols built to work with firewalls.

Networking

Applications are now thought of not in isolation, but in term of networking.

Network knowledge is important. Not in terms of knowing the names of the protocols, but in terms of the logic behind the thinking.

Network effect => there are more viruses for PCs, because it is a lot more than macs.

Bluetooth used to cheat in exams.

Hence, the challenges are not due to the limitations of the model chosen to design information systems. It is also limited by the choice and availability of technologies. We often think that technologies are not important. Yet we are limited by the programming languages, the standards, etc. We need to know technology to understand how it affects the design of information systems.  However, ISORG is not a course on computer science.

Data Model

IS Design is about how information is produced, exchanged and processed in the organisation.

The DM is used to identify the way in which data is produced, exchanged and processed in the organisation.  Look at the data processes and describing at the highest possible level the data processes.  We can then identify if there are steps which can be improved, and if there is a possibility of designing information system that can do this more efficiently.

The representation of the organisation is a very complex data flow diagram. Identify the areas where the data flow can be improved. This means that the organisation can be completely abstracted by the data flow diagrams.

Space of places => space of flows.  Space of flows is a simple example of data flows depicting exchanges of data.  We are not changing our business.  Just making smoother the way the data is being exchanged to provide quicker access to data.  This improves the overall data efficiency of the organisation.

The simplicity of the DM stems from the focus on the data, and not the organisation. By digitising data, we can achieve a more efficient workflow.  By focusing on the DM, the representation of these dataflows is usually very clear. No ambiguity.

Example. Automated class attendance taking system. Process is the same.  People come to class and take their attendance. Just more efficient.

However, many things are ignored. We need to understand the limitations of the model, not because one model is better than the other. One may be richer than the other. If you understand the limitations, you understand when it should be applied.

DM is not just 1 approach. There are different approaches. The most famous of these approaches such as BPR. Intent is to identify the data flow and to re-engineer/re-design the organisation by providing a better and more effective data flow.

Example.  Judge story.  The judge has to write the sentence which is then typed out by a clerk who would bring it to the recording officer to be entered into the system.  The manual process took 2 months.  To speed up things, it was decided that the judge should just type out the sentence himself and send it by email to the recording officer who need to input the sentence the system.  So this shortens the chain from:
Judge -> secretary -> Clerk -> secretary -> Recording officer -> Archives, to
Judge -> Recording officer -> Archives
Problem was that the wife of the judge left for the recording officer 3 weeks before the new system was introduced.  So the judge refused to send email or have any contact with the recording officer. This made the new system worse than before.  To correct the problem, they have to reintroduce the clerk to intermediate between the 2:
Judge -> Clerk -> Recording officer -> Archives

The adoption of the same system does NOT lead to the same outcome for different organisations!  Technology is NOT enough to lead and drive a specific outcome. What is important is the relationship between the technology and its users. So we need a richer analytical model than the one provided by the DM.

Eg. of choice to go to KL fastest may seem to use the Ferrari vs Hyundai. However it requires a trained driver to drive the Ferrari. Hence the optimal solution is not defined by the technology, but by the ability of the users to use the technology.

Tabula rasa = The organisation will do whatever is needed to use the technology.

Use of examples => make no assumption that markers know examples given by Antonio.

Relationship between BPR and DM

BPR can be conceived as 

a. an approach to designing the organisation in line with the data flow which occurs in the different processes. The intent is to redesign or re-engineer the organisation from a functional structure to a process-driven one.
b. an approach to encourage the adoption of information systems.  ICT is very important for this because it can be used to streamline the processes representing these flows.

BPR is a plan to change the way the organisation works. So need to know how the organisation works, => use DM. Then decide how to reengineer (ie. use ERP).  The information systems (eg. ERP) implemented will then change the way the organisation works., fulfilling the intent of BPR.

BPR and ERP

BPR is the data model behind the design of ERP systems. ERP are systems designed to optimise the data flows within standard processes in the organisation. ERP vendors assume that every organisation undertakes very similar processes.  [WT: So ERP can be "pre-packaged software"]. They do procurement, take care of payslips, etc.  We can design information systems which optimises these processes because they have the same requirements.

When we buy and use an ERP, we have to change the way we work according to what the ERP recommends.  Hence, ERP helps the BPR process.  The information system can be used to inform and support and lead the changes in the organisation.

Hence, while BPR is an approach to inform information systems design, ERP can also trigger structural changes in the organisation which optimises the data flow.

However even if the information systems are efficient, it may not be as effective compared to even some manual processes.  Eg. attendence taking by manual name calling and marking is more accurate than the automated system which may be subverted by students.

Decision-Making Model

Richer model because it looks at the organisation from a more complex viewpoint. Not just looking at the data flow. The main activity is to improve the way decisions are made by seeing the organisation as a bundle of decisions. So we need to understand how decisions are made.

What is a decision? Decisions can be seen as a rational selection amongst alternatives.

Rational decision-making:
1. The Goal must be clear (Go for dinner)
2. The Alternatives can be listed (All restaurants in Singapore)
3. The Criteria is unambiguous (Criteria: Good quality for $$)
4. The Choice can be made (Pizza hut) 

The choice cannot be made if the criteria is unclear.  Criteria need not be solely on $$.  It could be based on a combination of cost, ambience, preference for certain types of food, nature of event, etc.

In rational decision-making, there is no uncertainty in any of the stages of decision-making.  However, this is not always the case.

For example, we may not be always certain about the GOALS.

Example. Why are you at the revision today?  This is not a rational decision. The goals are not clear. Very difficult to get one answer.  Eg. if the goal is to get the highest marks, the alternatives could be to attend lectures, to study at home, to read widely, to talk to friends, etc, and not just to attend the revision class.

In the rational decision model, every stage in the decision making process is clear.  If the goal is not clear, alternatives are not always known, the criteria may be conflicted, then we are boundedly-rational. Cannot make optimal decisions.

Example.  Tic-Tac-Toe. Goal is very clear. Alternatives are known. Criteria is clear - win the game. So no uncertainty. So we can make rational decisions.

Chess. Goal is very clear. However, alternatives are unknown (too complex). Criteria is clear. So bounded rationality.

For information systems to make decisions, there cannot be uncertainty.  If there is uncertainty, we cannot create programs that can automate decision-making. But we can design information systems that can help us make better informed decisions. Eg. systems that can help predict weather conditions. May not be completely accurate, but we can still make better informed decisions.

Systems that are made to make decisions, vs systems designed to support decision making are very different. So it is 2 clusters of systems which do different things. One is intended to automate decision-making, and the other to support decision-making. The way the systems are used to collect and process information is very different.

Example. Transitlink card information systems that control access to MRT stations.  Does it make decision? Yes. The system has to make decision on whether to let you in or not by determining the amount of credit in the card.

In any example you use, you MUST explain why the decision is made. There is a difference between collection of data and decision-making examples. Attendance taking is NOT decision-making. Data collection is DM. So attendance taking is DM. [WT. Security doors, MRT stations, ATMs, are all examples of transaction processing systems that make structured decisions on behalf of humans.]

Decision making in organisations

We need to identify areas of the organisation which are affected by rational decision-making and those that are affected by bounded rational decisions.

3 levels: Structured decisions which are decision-making that is not affected by uncertainty. Semi-structured and unstructured decision are those that are affected by uncertainty and bounded rationality to various degrees.

Low-level decision-making.  Here, people are making decisions which are not affect by uncertainty. Problems are known and routine.  Solutions are known. This is the best place to introduce technology. No errors will occur.  In the 1970s and 1980s, most information systems have been introduced at the lower levels of the organisation.  Eg. assembly lines require structured decisions.  So robots are introduced because they don't get tired and the can make optimal decisions all the time.

Semi-structured/unstructured decision-making: Information systems are not used to automate the decision-making.  They are primarily used to supply information. Eg. stock brokers and traders' decisions are unstructured.  We cannot use information systems to automate these decisions.  However, traders use a lot of computers, not to make decisions for them, but to get more information that may help to improve the quality of their decisions.  They will not know if their decisions are optimal or not, but by getting access to real-time information, they can make better decisions.

What type of information systems can we use to support situations A (PK-SU) and B (PU-SK)?

Example.  An MRI scan of a patient does not give the answer to the medical ailment.  However, it provides information that helps the doctor to understand the nature of the problem.  The doctor will then be able to choose the right solution.  There is no uncertainty in the choice of the drugs to use for each symptom.  But the doctor may not know if the drugs exist, so information systems can be set up to help the doctor to quickly identify these drugs.  So systems can be used to support the "known" part of the problem, and other systems to reduce the uncertainty of the "unknown" part of the problem.

Example. ISORG is based on critical assessment of the use of technology in organisations.  So the exam questions are unstructured - more than 1 way of answering the question, no real right or wrong answer since it is not possible to find an "optimal" answer.  Some students may argue that the answer to questions is "known", although the actual exam questions are "unknown".  But most times, the solutions are specific to to the questions, so it will be wrong for students to prepare "structured" answers for the examination.

How do we design an information systems for problems with high uncertainty?
1. Should we design the IS to provide as much information as needed about the subjects that people know less about, ie. focus on areas with more uncertainty
2. or should we design the IS focusing on what people already know, and fill the gaps where there is less uncertainty.

Example. Students taking ISORG. At beginning of the course, knowledge is zero. Is it better to learn by giving full details and information about the subject which students know nothing about, or to provide only enough information to build on what students already know to first build their fundamentals?

Antonio talked about the "competency-decison gap".  What makes sense is to give more information only to people with a certain level of competency.  Otherwise, the less competent people will end up with information overload => simply because they cannot process the new information quickly or effectively enough.

Example. Formula 1 cars have 38 buttons on their steering wheel to control different functions.  Untrained drivers cannot drive such cars because they lack competency.  So giving them such a car to drive will reduce the level of their driving, and they can probably drive faster in a normal Hyundai car.

The big challenge is to find out the right balance between information flow and the competency of people who need to process the information.  Decision-makers are limited by their bounded rationality => unable to process all available information.  From a strategic viewpoint, it is better to give more information to people who already knows how to process the information.

Example: In the 1970s, IBM and Xerox dominated the IT industry.  They realised computers are good businesses in the office machines industry.  Their approach to designing informations took 2 very different directions.

IBM got a group of engineers to look at how people work in organisations to try to design systems for them. The engineers focused on structured decisions.  So they designed information systems that are effectively Office Automation Systems (OAS) and many other applications which we still use today.  Microsoft Office has its roots from this process, and is designed to help the production and reproduction of documents.

Xerox PARC, on the other hand, employed anthropologists to study the organisations. These anthropologists merely observe, without judgement. They wander around to understand what the people did, to answer: "What do people do in the office?"

In 2 weeks, they reported that the main office activity of people: they chat all day long.  So the next question is what are they chatting about?  So they put tape recorders all around (within the limits of privacy regulations).  They found that 95% of conversation is not related to work.  Mon, Tue they talk about what they did in the weekend.  Wednesday they complained about their colleagues.  Thu, Fri they talk about what they plan do in the upcoming weekend.  Most conversations are held around water-coolers.

The remaining 5% of the conversation is about clarifying what needs to be done in the work.  This is strange because there already exists written rules.  What was discovered is that the only way the organisation works is not going by the book, but rather by adapting rules to the organisation's culture and people.  The 95% of conversations were essentially used to facilitate the 5% of discussions.

So Xerox focused on facilitating intra- and inter-company communications.  They invented Ethernet (the local area network protocol widely in use today), and many Internet protocols. They invented the Alto, which is essentially the world's first mouse-driven PC.  They invented the WIMP GUI interface, which is the underlying technology used in Mac OS and Windows.  Apple took some of the Xerox PARC engineers to design the first Mac. They created Apple Talk which is a very successful LAN. It was designed to support communications.

[WT: Though Xerox PARC is commercially a disaster because the management was unable to understand how to exploit the technology they invented, the important lesson learnt here is how a proper understanding of the organisation culture with respect to decision-making can lead to design of information systems that accurately relates to users' real needs.]

In 1995, Microsoft implement Win95, but it was not designed to support communications and networking.  Microsoft used the IBM approach to study organisations. MS Word does not allow collaboration in documentation.

The DMM is richer than the DM. DM only relates to the streamlining of the data flows. DMM can automate and support decision-making. So information systems designed are better tuned to the organisation. But it will not fully eliminate problems.

Transaction-cost Model

The organisation is described as a network of transactions.  We need to understand the nature of transactions, such as the rules governing transactions.  By looking at transactions, we note that most problems have to do with poor access to information.  So the question is how we can design information systems that can improve the access to this information.

Transaction costs are frictions.  They are ancillary costs needed to make the transaction possible.  If transaction costs are so high that they prevent transactions from taking place, we have a market failure.

Organisation => Model => Network of contracts => Information Systems => Organisation

TCM is based on a very strict assumption: that economic agents will always act to maximise the outcome of their exchange.  In doing so, they will add to costs to protect their self-interest.  These costs represent transaction-costs.

Microeconomics theory states that price and quantity of goods exchanged is based on the equilibrium between supply and demand.  However, it is not always possible to know the true relationship between the price and the quantity of goods.  While we obviously want to know the minimum price for the maximum quantity, the optimal price is not optimal in absolute.  Every individual has a different preference. They have different perceptions of what the optimal price is. Agents have different cost to find out what is the optimal price

Uncertainty is a major contributor to price fluctuations.  Example:  The cost of rice production is affected by weather which cannot be predicted, so it is not possible to predict the exact price of rice.  In fact, the most powerful computers in the world are used heavily for weather prediction, give the huge geo-economic implications that weather has. Accurate predictions can reduce uncertainty.

Another example: petroleum price is determined by unpredictale geopolitical situations, hence it is not possible to know the price.

So there are lots of environmental uncertainty.  Because we want to maximise our value, we invest in systems and processes to help us maximise our resources.  This constitutes transaction costs.

There are situations where uncertainty is not due to unpredicable environmental factors, but due to bounded rationality.  Example: where to buy the cheapest pen?  Bounded rationality and uncertainty affects the relationship between Price and Quantity.  Hence we invest in economic resources to try to reduce uncertainty and bounded rationality. This represents transaction costs.

Opportunistic behavior:

Opportunism may lead people to cheat in order to maximise their benefits. It is not just the opportunistic behavior that increases costs, but also the fact that others may feel that they have to protect themselves from such opportunistic behavior, leading them to invest in systems to protect themselves.  These systems are transaction costs.

Bounded rationality:

Opportunistic behavior is created when knowledge is uncertain or incomplete. Example: Buying a 2nd hand car.  If we are uncertain about the quality of the car and we don't trust the dealer, then we may need to hire and independent professional to help us evaluate the car or regulate the transaction. This increases transaction costs.

In a perfect world with perfect knowledge, there is no opportunism.  Tourists get robbed all over world because they do not know the real price at new places that they visit.  When there is no uncertainty, opportunistic behavior has no effect.  In perfect market with perfect knowledge, everyone will be trading at the same price and quantity.

Small numbers:

Yet even with perfect knowledge, people may still be victimised by opportunistic behaviors.

Example: If hairdressors have the same skills, present everywhere, and offers the same prices for standard services, how would you choose your hairdresser?  The "rational" choice would be the hairdresser nearest to you.  However, after one year, if you are used to the local hairdressor, you might not want change the hairdressor even if the hairdressor increases price.  This is because there is "lock-in" effect where there is switching-costs of having to teach the new hairdressor how you want your hair done, travelling costs, etc.  So you may still stick to the hairdressor despite the opportunistic behavior.

Hence even if at the beginning, there is an infinite number of suppliers, after a certain time, you will encounter a small numbers situations.

Example: Ferrari produced F4 and wanted a unique windshield so that the driver will always have the same contrast. They looked for suppliers and found one who can supply this. They found the supplier and was finalising the contract. Ferrari wanted to add one line to the contract to force the supplier to license his technology to the supplier's biggest competitor.  This was of course against the interest of the supplier.  The deal was never signed.  Ferrari was afraid to be subject to seller-side small numbers opportunism.  This is a classic example of a market failure. The transaction did not take place because of bounded rationality and uncertainty. Transaction-costs make it impossible to carry out the transaction. 

Firms vs Markets

Can we reduce transaction-costs in this solution? Yes, by using a Firm.

Instead of investing in resources to manage the transaction, we can use a firm. The firm has norms and rules which are respected.  Firms use employment contracts to control behavior.  There is a big different between employment and project contracts.  As employees, we don't get paid for everything we do for the firm.

Example. Lecturer is paid for a certain number of hours. Students pay for sitting in the room for a certain number of hours. Lecturer can say what he wants. No specific requirements. So the best approach is to pay or get paid depending on the quality of the session. However, very hard to determine criteria. Students may be opportunistic because they do not want to pay. So lecturer cannot accept that the measurement is done by students. So there is market failure.

Hence we need a third party to select the lecturers based on their competency and commitment. We have to create a situation with a common interest - lecturer wants to give the best lectures, and students want to get the best out of it. This process is complex and adds a lot of transaction costs - to identify the agent who will not be opportunistic in the transaction.

So a hierarchical relationship allows us to select agents in a different way from how we might choose agents in a market form of organising.  The key criteria is goal congruence.  Eg. professor in university has to do R&D which involves very high task uncertainty.  Very difficult to come up with standard ways of measuring progress.  Hence the university must invest in a lot of time and effort in interviewing and evaluating the person for the job.

In hierarchies, we invest in costs of selecting the right person in order to eliminate the possibility of opportunism.

Search costs

This relates to the problem of buyers looking for sellers, and for sellers to look for buyers.
Example. Companies invest in marketing to reduce search costs.
Example. Why is the rent for a shop in a place where many people pass by higher than the rent for a shop where people don't pass by? It is to reduce search costs. Pay more money to get access to more customers.
Example. SLS can be described as an institution that attempts to reduce search costs by putting all the shops together.  Rent in SLS is higher on the first floor than the higher floors, and shops at the back. Shops at the back is more for techies who need specific parts, not from the general market.  Customers of back shops know what they are looking for, so no point for these shops to pay premium rent for the front row.

Factors that affect search costs:
- Uncertainty/complexity
- Bounded rationality
- Antonio stressed that opportunism does not affect search costs.

Contracting costs

Search cost should not be confused with contracting cost. You have found the ipod, now you have to find the price to pay for it.  Contracting costs is trying to find the right price (price-qty) for the product.  When quoting the SLS example, we must understand that "looking for the right price" is NOT search costs, but is actually contracting costs.  Search costs is only incurred in looking for potential agents to transact with.

Factors affecting contracting costs:
- Uncertainty/complexity
- Bounded rationality
- Opportunism (seller will try to sell you the worst product, buyer will try to pay the least price)

Control/Regulate costs

Ensure that we get what we paid for.

Factors affecting control costs:
- Uncertainty/complexity
- Bounded rationality
- Opportunism

Note: Small numbers are left out because it is normally not part of the transaction life cycle.

So how to design information systems to reduce transaction costs?

Reduction of Search costs:

Any system that can reduce the cost to find potential buyers or sellers. Some of the most successful businesses today are where they are because of their abiility to reduce search costs. Example. Facebook ($60b valuation) is successful due to its ability to reduce search costs. It allows businesses to do market segmentation by focusing on customers with characteristics that they wish to target.

Main business of marketing companies today is search engine optimisation (SEO). If you want to be found on the web, you must be searchable by search engines.  Previously, it was Yellow Pages.

The Internet without search engines is worth nothing. Companies used to spent huge amounts of money to buy URLs just so people can remember these names easily.  They are not doing it anymore. They are now paying large amounts of money to marketing companies to promote their presence on search engines. Eg. Pizza companies pay lots of money to get listed at the front of the search results. Either you are in the first page or you are out.  Top part of search page is PAID services.  This is for companies who have not invested in SEO.  They simply pay Google to put place their banners up front.  Google must guarantee that only the people interested in the product will actually see it.

Search engines will also segment by IP addresses to make the search more relevant and reduce search costs further.

Other examples: Ebay try to match in the best possible way, the interests of buyers and sellers by reducing bounded rationality and uncertainty/complexities. Asiarooms reduce search costs. You have found the room, but if you want to find if the room is worth the money, it is contracting costs.

Google is top search engine, and must keep giving the best results. Bing is trying to be like Google. Search engines must continuously optimize their search results.

Reduction of Contracting Costs:

Tripadvisor.com is good for reducing contracting costs by reducing opportunistic behavior. It provides you the best information about the hotels. It also tell what kind of customers has provided the reviews (eg. Family vs individual needs). It is a clear system designed to reduce opportunistic behavior of hotel owners.

SeatGuru.com provides the best seat for the plane. Seats close to toilet, good legroom, etc. Just making more transparent what I get charged.

Any system that helps better understanding of the quality of the product we are getting will help reduce contracting costs.

Reduction of Control Costs:

Tripadvisor.com also allows you to book the room. Tripadvisor.com is investing a lot of money to protect the integrity of users' reviews. They have to protect themselves to prevent hotels from writing good reviews themselves. So they have to invest in IS to protect themselves.

Shopkeepers prefer credit cards than checks despite the transaction costs. This is because the credit card company will pay the shopkeeper (guaranteed). Same thing Paypal is doing.

Ebay mainly reduce search costs. Ebay created a rating system of buyers and buyers to reduce control costs. It helps to reduce risk of opportunistic behavior which can affect the actual confirmation of the transaction. Ebay also requires insurance? Reduces opportunistic behavior of sellers.

Example: RFID to tag freshness of food. So technology is used to track the food. Mad cow tracking of beef. 

Post-mortem of Mock Exam

Quick comments

• Generally "not bad".
• Examples should be your own.
• Examples should match what you are talking about. A wrong example is worse than no example.
• Must answer the question!
• Each question is
• Q3 and 4 are the core subject matter
• Q1,2,5 and 6 are "specialist" questions that add an extra layer on top of the core subject matter.  You need to discuss how information systems are used in competitive environment in the context of the analytical models. 
• You should make a sketch of the structure of what you want to write about before you start writing.  After making the sketch, think about the best question that your sketch is providing an answer for, and then see how much the question matches the actual exam question.  That will tell you if you are answering the question.

Question 1: Why the choice to invest in electronic business is not always a success?  What are the conditions that make it successful?  Theoretically justify your answer.

In ISORG, for all purpose, e-business = e-commerce.  Hence you must talk about e-commerce when you see the term e-business.

Theorectical justification must use one of the 3 analytical models.  Antonio prefers the TCM for this question.  You must make the bridge between e-business and transaction costs.

Distinguish transaction costs as something that we pay on top of the price, ie. the real cost of exchange is Final Price = Price of good + Transaction Costs.

Why is a company like Megabytes (SIM canteen operator) not doing e-business?  That is because their search, contract, and control costs are all very low already.  Search costs are low because they only serve students and faculty within a specific locality.  Contract costs are low since they collect payment on delivery of the food, and customers cannot negotiate on prices.  Control costs are very low as there is no need to monitor performance after the food has been given to customers.  Can these costs be further reduced using information systems, such as allowing students to order online?  Not at all.  The additional cost of implementing and operating the system are transaction costs that only serve to increase the overall operating cost for Megabites with no increase in revenue.

Good examples: Amazon.com is successful because it is able to cut the cost of reaching out to customers and servicing its customers using the Internet while disintermediating the brick and mortar retailers and distributors.  Apple iTunes disintermediated the physical need for sending a CD to users, allowing users to pay for and consume the music immediately without having to wait for a CD.

We must remember that transaction costs are different for different people.  For a customer who have no access to a local store that sells the right-sized clothes and hence have to import the clothes from an overseas supplier, an e-commerce site that offers prices that are lower than what he would have normally incurred through his regular channel will be very attractive.  Yet for another customer who stays near a physical store offering the same clothes, the e-commerce site is not attractive because of the relatively higher cost of buying from the virtual store.

When giving examples, students must avoid using clichés without explaining them.  For example, if you want to talk about how information systems can be used to control the freshness of fruits and vegetables, you must talk about the use of RFID to track pallets, which adds to control costs.

Question 2: Information systems can be conceived as strategic weapons.  Explain under which circumstances an information system can be considered a strategic asset for a company.  Critically discuss with examples.

You must first explain what are strategic information systems.  They are systems that allow companies to distinguish themselves from competitors, and make their products or services unique offerings to customers.  Technology does NOT provide long lasting competitive advantages.  For example, the iPhone used to be unique as a piece of hardware, but it no longer holds that competitive advantage as other manufacturers have been able to effectively build better hardware to compete.  The iPhone's key competitive advantage is now the App Store, which is unique to Apple.  However, even this will not last forever.  Android Market is catching up quickly.

Hence if you use American Airlines' SABRE to talk about how it enabled AA to compete effectively in the airline industry, you must realize that online airline reservations systems have become a "standard", a necessary element for any airline to possess in order to operate in the industry.  So you cannot consider such as system as providing competitve advantages anymore.

Technology is easily copied.  What makes technology competitive is the way it is used by different organisations.

Question 3: Critically discuss the effects of IS on search costs.  Discuss, with examples, if IS can reduce these transaction costs.

This question is on search costs only.  Do not overdo discussions on costs in other stages of the transaction life cycle.  Do NOT talk about opportunistic behavior, because search costs are not affected by opportunistic behavior [WT. Note that you actually can argue that search costs are affected by opportunistic behavior if you use the example of how a potential seller is UNLIKELY to tell you who his biggest competitors are (because of opportunism on the part of the seller), and so you have to incur cost of searching and doing the due diligence yourself.  This can also be discussed in terms of information asymetry, which is related to bounded rationality].

Focus on bounded rationality (information asymmetry, information overload) and uncertainty/complexity.  Discuss how technology can be used to reduce bounded rationality and uncertainty/complexity.

Sim Lim Square story (ie. user goes around the building trying to find the best price for a specific HDD that he wants to buy) is not a good example because it relates more to contracting costs than search costs.

Critically discuss the limitations of using technologies.  Examples: Search engines must be constantly scanning the Internet for updates to web content.  Auction sites like Ebay must find ways of making it attractive for buyers and sellers to transact with each other on the platform.  Social networking sites must constantly find ways of keeping its users and generating network externalities.

Question 4: What is decision making?  Explain, with examples, the design of information systems that support structured decision-making.  Critically discuss the limitations of this approach to IS design.

ATQ = Answer the question.  Focus only on structured decision-making.  While you can write about unstructured decision-making to distinguish structured decision-making, do not over-write on it.

Examples should be in support of your points.  It is not the quantity that counts.  Do not be afraid to invent examples to fit your needs.  Example: If you use the class attendance system as an example of structured decision making, you MUST explain what type of decisions it helps to automate (eg. the data collected is used to determine whether a student is allowed to sit for exams, and for the Ministry to find out if foreign students are really studying the course instead of using their student visas for other purposes).  Simply mentioning the class attendance system will NOT earn you credit, because it could be simply seen as a data collection system and hence is better designed using the DM instead of the DMM.

To answer this question, explain the decision making steps (Goals->Alternatives->Criteria->Choice).  Then focus on explaining the structured decision making process.  Bring out the point that structured decisions are AUTOMATED, and not simply "supported".

[WT: Good examples of TPS include ATMs (because the machine has to decide if it should dispense the money you requested), IVRS (interactive voice response systems, aka voice answering services used by companies like banks, because the system has to provide menus of options based on the caller's selections), security door entry systems (because the system has to decide if the credentials presented by users are authentic and if the user has the right access level to enter the room).]

Limitations:

- DMM ignores opportunism.  Even for structured decision, opportunism can come into play.  When people's jobs are threatened by automation, they will try to stop or sabotage the implementation of information systems that can replace them or make them "lose power".  Example: Passport application in certain countries.  Even though the process involves structured decisions, the administrator processing the application wields power and can control the time it takes to complete the task.  Knowing the administrator and giving "benefits" to the person can help to "speed up" the process.  If the system is replaced by an information system with transparency into the entire process, these administrators lose their "power".

- The system has to be flexible to react and adapt to changes in the environment.  Too often, the business rules are hard-coded into the system, making it difficult and costly to change in the future.

- There is still a need for humans to handle exceptions even when key structured decisions have been automated and made by machines.  Example: At immigration checkpoints, we can now go through automated gates simply by scanning our passports and our fingerprints.  This speeds up processing of "good" cases most of the time.  However, if there are issues, eg. the system is unable to verify the credentials because of passport expiry, or wrong user actions, the system will simply refuse to open the gate and alert human administrators to come over to rectify the situation.

Question 5: What is an information systems strategy?  Critically discuss the role of technology in defining the strategic implications of ICT solutions.

You have to distinguish between ISS and IT Strategy.  ISS links what you want to do to the design of the technology.  Can use the Porter's model.  But Porter himself noted that technology does not give lasting competitive advantages.

Question 6: Information systems can disintermediate the value chain.  Explain how and when this can happen.  Critically discuss with examples.

You must used the TCM to explain this.  Cannot use the DM or DMM!

Example: Barnes and Noble.  Why do we need wholesales and retailers?  Intermediaries are there because they get paid to facilitate transactions.  They exist as long as they add value to meeting the customers' needs.  Today, with e-books, we have an even higher degree of disintermediation.  Authors can publish their own books without a publisher.  Books get bought and consumed by customers without the need for physical shipping.  Logistics companies such as Fedex and UPS are definitely not happy with e-books, because they have been disintermediated from the process.  New Internet intermediaries such as credit cards and payment services like Paypal are happy because they provide value in the e-commerce value chain.

Not all services can be disintermediated.  For example, car manufacturers still need to have country-specific distributors to service local customers.  Local retailers provide additional services that are difficult for manufacturers to do.

You need to discuss all the above points in relation to the use of information systems and technologies.