Not much time to write (the second post about the gold market is already on the making), but an intense week overall, as always.
Internet Explorer is a bad browser. Period. Spent several hours this week fixing problems related with the box model on Internet Explorer. For example, CSS shadows act very differently on IE and on Chrome of Firefox, potentially breaking the text by overflowing on the top of the box. This can lead to text that appears to be “cut”, specially when using diacritics with tall fonts. I guess that, at some point, we should try to push that no browser comes preinstalled with any operating system.
I have made some quick notes on experiment design. They can be resumed in this few points:
- Go from particular to general. This way it is easier to catch up with the different probabilities of the happenings, specially if you have things that need to happen equiprobably.
- Go thoroughly through the numbers several times. It is easy to overlook things.
- Write a script of the experiment that explain how will you interact with the subject, in order to avoid the introduction of biases due to your will of the experiment going well.
- Write down the protocol. Specify the number of repetitions and calculate the approximate length. Check if the length can affect the results that can be obtained.
- If you are evaluating task execution, take always into account the learning effect.
- We all love our babies, but sometimes they don’t give us the results that we really desire. A failed experiment is a great opportunity to learn, not lost time.
- Data grabbing is as important as data analysis. Both of them can wreck the results and render a whole paper invalid. Give both of them the needed importance and don’t relax the protocols.
- If you think you need it, use the services of the statistics consultants of your University if you have one.
I have tried to enforce all this points in the experiment I am currently working on.
The Value Of Money
I knew already, but this week somebody taught me again that money loses value with time. Money today is less money tomorrow.
Never underestimate the power of noise to mess up with your health, specially sleep. I am forced to live for some days into a quite noisy environment and I am already noticing how it does affect my health.
We have been talking about new ideas in my research group. It is always refreshing to see that we have a lot of stuff to do and a lot of ideas and proposals to bring to reality. Maybe at some point I can tell something more…
I still keep on thinking that us, the computer scientists should do a better and bigger work on divulgation. If you have any ideas, just tell me. I am open to collaboration. Just write a comment here and we can start from there.
HTML5 RANSAC Visualization
And I guess that this is all for this week.