In this assignment, you will write a position paper. You must choose from from the possible position topics, or clear your topic idea with the professor. Note that you can also take the opposite position from those described below.
You will also evaluate one of your classmate's position papers, and then revise your paper. Your evaluations should be about 1 page each.
Above all, try to have some fun with your position --- invoke some reaction in your classmates (from awe at your genius to anger at your toeing the party-line).
First Draft Position paper: Friday, March 13th, 11:55PM
Paper Reviews: Friday, April 3rd, 11:55PM
Revised paper: Friday, May 8th, 11:55PM
Upload your papers in the Sakai Assignments page.
People own their generated data. Each data, people generate data on many aspects of their lives: their location, who they communication with, their shopping, and their search terms. This position argues that ownership of the generated data lies not with the technology platform, but rather with the end user. Users could demand payment, sign away their rights, but this position argues it is the user who owns the original data. The counter position is that the technology creators should own the data.
Net Neutrality enabling a wealth of new services and business As Internet Service Providers are "common carriers" --- in return for their limited liability of the traffic they carry, they must not discriminate based on content of the traffic. This insures a fair an efficient market for business to operate, as communication is an essential function for all businesses. Without net neutrality rules, companies, political organization, or other groups could be 'unplugged' by ISPs at any time. For example, innovative local network provides could be have their traffic shut down at any time.
Net Neutrality stifles innovation Today's ISP, have an obligation to maximize shareholder value, and traffic discrimination is one tool to maximize their return on investment (ROI). By maximizing their ROI, ISPs will be able to invest in new products and services, while net neutrality rules will only limit their profitability and innovation. For example, an ISP could offer free service to low-income populations using and ad-supported model. a
CPUs will retain their role as performing the dominant amount of computation for Internet Services. Despite progress on accelerators, for example, TPU, GPU or FPGA, the CPU will perform the bulk of the computing needed for Internet Services.
CPUs role in computing will substantially diminish. GPUs FPGAs, and specialized hardware will become cheaper and easier to program. There performance inprovements in latency, throughput and energy will reduce the CPU to a manager/housekeeping role, rather than perform the build of the computation.
Internet connectivity will fragment in a manner visible to end-users. Due to security and control issues, the Internet will fragment such then end-to-end connectivity will not be consistent depending on the attachnment point. For example, connecting to services inside or outside China is widely different. Similar effects will happen for work or home use, as well as to various services (e.g. video sites such as Netfix) depending on if you are at your home or on a different device.
The uniform access modle of the Internet will stay mostly intact. The Internet was designed mostly to allow peer-to-peer exchange of data. In spite of some hiccups, and one or two notable exceptions, the flat model will remain. Protocols such as STUN and TURN will enable peer-to-peer connectivity in spite of a diverse set of middle-boxes.
Voice Interfaces will eclipse touch screens as means of accessing services.. Today most people still use screens and touch interfaces to interact with web services. However, services like calendars, email, shopping and and general search will move to where voice interfaces dominate.
Trust will significantly undermine the use of cloud services. Lack of trust can arise for the following reasons: (1) ability to secure data from both inside and outside attacks, (2) raising prices resulting from monopoly lock-in, (3)serious outages. Due to the able trust issues, many large-scale service users to manage their own clouds. This will mean more organizations will run their own services or clouds on their own infrastructure.
Technology Platforms will Democratize Education. Traditional education has been limted to the well-off, both in terms of money and socal networks. However, new educational technologies, such as Massively-Open-Online Courses (MOOCS), combined with cheap access to content, through technolgies such as Chromebooks or the One-Laptop-Per Child, will democratize education such that any child can have a world-class educational experience.
Educational Technology Platforms are Detrimental to Learning. New technologies that promise to revelutionalize educations, including MOOCS, Chromebooks, One-Laptop-Per-Child, are all hype. The value of education lies in the human interactions: including teachers, peers-leaders, and coaches and councilors. Without access to human resources, the educational experience of students will be poor.
The vast majority of large-scale Internet Services will use hosted cloud solutions. The efficiencys due to economies of scale and statistical multiplexing of users, and risk transference outweigh any potential negatives for running services on 3-party clouds providers. undermine the use of cloud services. Lack of trust can
The current username/password model is fundamentally insecure. New models of authentication are required in order to keep attackers from creating a substantial drag on usability and applicability of Internet services. Such models include biometrics and 2-factor authentication schemes, and novel multi-role authentication paradigms.
The Username/password model may have limits, but these are easily fixed through straightforward extensions such as enforcement of strong passwords and password managers. Other authentication models are too burdensome and counter-intuitive for use in practice.
Although system designers often reference elegance and consistency, backwards compatibility, and portability, and extensiblility trump those metrics for many real world systems and development environments.
Generality, simplicity, speed and reliability are at odds. Too do one well requires sacrificing some of the others, and these trade-offs are fundamental.
There always exists and elegant design for a problem that is simple, functional, fast and reliable. If a system appears to sacrifice one of those metrics for another, you have not worked hard enough or smart enough.
While virtual machines provide strong isolation guarantees, in reality containers provide most of the isolation needed to run Internet Service applications at a fraction of the resource usage of virtual machines.
Reliability is proportional to service complexity. Very reliable systems do a few things well. This position argues that it is reliability is proportional to complexity, and thus complex services will always have low availabilities (e.g. 3 nines or less).
Automation will solve reliability problems. Both software bugs an operator errors account for the majority of downtime. This position argues that by providing automated tools, a high degree of availability can be obtained (99.99% or higher). before copyright law.
This
paper has some good guidelines in for position papers in general.
Two sample “good” position papers from past courses can be
found here
and here.
First, make sure to articulate your position clearly. Second, for a good computer science paper, you should have some quantitative arguments. A list of anecdotes is not a persuasive way to support of a position. Sometimes, you can't directly measure something, but an indirect observation might support your argument. For example, some people have made the argument that performance isn't as important as it used to be because the difference between the average selling price of a PC and the most expensive PC have diverged over time. While not proving the argument, the thesis fits the facts better than many alternative explanations.
Third, be careful of using counter-examples to argue against a position. For example, a position of the form "X implies Y" and then coming up with an example of "not Y" doesn't say anything about statement X. Counter examples can be quite useful, but make sure the position is clear enough that the counter-example is meaningful.
Please us the following evaluation scheme on a scale from 1-10, where 10 is the highest quality and 1 is the lowest:
10. Excellent: The paper could be submitted as a "letter" --- a short position paper-- to a journal as is.
8-9. Very Good: The paper has some problems, but nothing that couldn't be fixed without a quick clean-up.
6-7. Good: The paper has some problems, there are some gaps in the overall positions, counter-positions, or supporting evidence.
4-5. Fair: The paper has more serious problems. These may include (1) ill-defined position, (2) elements of the evidence are missing, (3) counter positions are not addressed, (4) really bad grammar, or (5) poor organization.
2-3. Poor: The position in not well explained or defined. The paper is confusing or internally inconsistent. The evidence is non-existent or very poor.
1. Atrocious: What a piece of junk! I'm surprised this was turned in at all.