Saturday, August 18, 2012

On open access in scientific publications

A recent article in "The scientist"  issued a blunt attack on open access in scientific publications, call it: predatory publishing.  This article claims that open access publications will not allow people to distinguish between "science" and "non-science", since it will not guarantee the quality of publications through the rigorous process of peer reviews, and then gives the main reason for objection --- "implications for tenure and promotion will be significant".    Some of the comments compared the arguments with those made against open source software.   

I think that there are some orthogonal issues here:  first -- I am for open access, actually every person today can put any paper on a website, and now all we need is to index it so it will be accessible to all.   Scientific publication is not different in principle from newspaper article, blog, TV show, book - in all of them we can find low quality and high quality examples.  

The review process is highly subjective and noisy,  and the amount of false positives and false negatives is quite high, it provides a quality control which is better than nothing, but it does not guarantee anything.  There can be some quality evaluation models, but this is a different discussion. 

I have written before about the fact that  the need to measure everything is taking over work in the corporate world,  this is also the case in academia,  in many cases universities count papers,  or use some other quantitative metrics, which fits the current publication system.    A new system will require to change this criteria which is an earthquake in the academic system (which needs earthquakes from time to time)...

I think that open access in scientific publication is happening and resistance is futile.....

Sensors as actuators

Phil Windley, the CTO of Kyntex, has reported recently in his Blog about "talking back to a thermostat".  Phil has constructed a thermostat as event producer,  The example he talks about is changing the display color, however one can think on even more meaningful actions like turning on and off the air-condition.  The thinking of sensors as vehicles for human awareness is not enough,  in our universe, there will be more sensors than people, and especially much more sensors than people who are able to look at sensor reading treat is as data (small or big) and make sense of this data.  We are more and more move to era of autonomic computing, the Mars rover "curiosity" is largely autonomic, since the speed of light barrier creates a time delay that makes it virtually impossible for remote control.   We'll see more and more sensors who will also serve as actuators, or smart sensors directly talking with smart actuators (e.g. thermostat to the air-condition
controller).    The required event processing  (filter, aggregation,  maybe even patterns that takes into consideration other sensors' input, like presence of people in the house)  can be distributed between the smart sensor and the smart actuators without any middleware.    More on embedded event processing - later.