Participants: Popescu Lucian, Airinei Daniel
PL: Reflections on trusting trust
Key Ideas:
- Trusting a computer system is based on the same principles as trusting another human being.
- There are 5 strategies for trusting: optimist, pessimist, centralised, investigation, transitivity [1]
- The stragety we used is dependant on the environment we try to apply the strategy in.
- Can we force a computer system to never lie about the operation is runs?
- At what abstraction level should the operations be formulated so that the computer can ensure us about their trustworthiness?
- If we can't force the system not to lie then what approach should we take?
Questions:
- Are trust issues greater/of more importance in computer science than other areas of day to day life?
- To what extent can you prove trustworthiness?
- To what extent do you need to prove turstworthiness?
References:
[1] https://eprints.soton.ac.uk/260029/1/ISWC04-OHara-final.pdf Fig.1
https://dl.acm.org/doi/pdf/10.1145/800001.811674
https://www.cs.cmu.edu/~rdriley/487/papers/Thompson_1984_ReflectionsonTrustingTrust.pdf
AD: The End to the Myth of Individual Programmer Productivity[1]
Questions and Observations:
- The fact that progammers can be tracked by their programming style is GDPR complaint?
- What implications does the above question have to open-source code?
- The fact that the same piece of information is repeated in a paper is not bad. This technique is used to help the reader have the necessary context.
- The graph in the paper (I forgot which one because I don't have the paper ATM) does not have measurement units noted.
- It would be great if we would implement this experiment to our courses to see where students perform better and where they perform worse.
- Does the fact that the students were asked to log their performance during the test affect the results?
- Why did they chose C developers for their test population? Choosing another language community would affect the results?
- Why is it important to have developers focused on a specific language? Why didn't they focus on general computer science problems?
- Why is the dataset they used for this experiment not open?
- Is there a continuation of this study where one company (or multiple companies) tried to apply those advices?