How to decide on Resource Share
From Unofficial BOINC Wiki
Contents
|
[edit] Introduction
Deciding on Resource Share is mainly a personal choice about the merits of different BOINC Powered Projects. You should consider running at least two projects so that if one project is down, your computer can still do useful work for the other Project. Even if you are really only interested in one BOINC Powered Project it is still advisable to setup at least one BOINC Powered Backup Project. Our recommendation is that you have at least two, preferably three or more BOINC Powered Projects.
The good news is that we have a pretty good selection of BOINC Powered Projects now, so, you have a number to choose from.
Some of the factors that can come into play in the selection of a BOINC Powered Project are similar to the ones used to define the Resource Share to be allocated. Some of these factors include:
- The Potential Impact of the Science Performed
- The Probability of That Impact
- The Actual Science
- The Timing of the Science
- The Utility of the Science Produced
- Safety First
- Personal Benefit
- Sense of Community
- Competition
- Personal Interest
- Visual Pleasure
- Political Signals
- Democratizing Science
[edit] Factors
Each section below explains the factors and offers some explanations how a various projects may, or may not, exhibit these characteristics. Note, we are not making value judgments here, and the examples are not meant to laud one project over another. Nor are these examples to be taken as the end word. In other words, you will still have to make a judgment based on your own research.
[edit] The Potential Impact of the Science Performed
Many people will be more interested in the Results than the science behind the project. This is based on those who care if a project uses scientifically advanced methods and if it is intended to produce a result that changes the life of many people or even civilization as we know it.
[edit] The Probability of That Impact
A result in some projects could be dramatic, but should the resources devoted to a task depend not only on the size of the effects but also the probabilities of such results?
[edit] The Actual Science
The purpose is to help the science, so which project is most likely to generate the most scientific papers in the most prestigious scientific journals?
[edit] The Timing of the Science
The number and quality of papers may be held up by a shortage of computing power in some projects but an excess of computing power probably will not result in that much of acceleration in the number or quality of papers. So avoiding projects with excessive crunching power seems sensible. It also seems sensible to avoid projects which will take many years to complete enough calculations to get anywhere near an answer or a scientific paper. Putting your crunching power to work where it is most likely to help.
[edit] The Utility of the Science Produced
Some science can be badly used. Is the use of the science going to be beneficial? Also, who owns the results of the computation? If the results are valuable, will they be freely available to the public or will they belong to a for-profit business?
[edit] Safety First
Safety first: choose a project which you trust. This includes trusting (a) The project team to ensure that their applications are free of bugs that could damage your computer or cause security problems. (b) The project to use proper security practices on their servers. (c) The project to appropriately limit its use your computer. (d) To produce and use results only for beneficial purposes.
[edit] Personal Benefit
Aiming for scientific benefit sounds good, but it is unlikely that one persons crunching power will make any difference. Therefore join a DC project according to what you get out of it. This could either be
[edit] Sense of Community
Join wherever you are most comfortable with the bulletin boards etc.
[edit] Competition
Personal preferences will dominate whether you want to compete in the largest DC project or whether you would prefer joining the newest Distributed Computing project.
[edit] Personal Interest
If a particular project is interesting.
[edit] Visual Pleasure
If a project provides a visually pleasing and interesting Screensaver or other visualization.
[edit] Error Free Operation
Error free operation. There is little point in choosing a project that is just going to be annoying requiring babysitting through some processes.
[edit] Political Signals
Though it is unlikely that one persons crunching power will make any difference to the science, there are other benefits. If large numbers of people crunch for a project, it sends a signal about the importance of that project to politicians who might then take measures to support the goal of the project.
[edit] Democratizing Science
In a similar way, should the public have some sort of control over the direction of science? If the public do not like GM crops, should public opinion create some sort of feedback control? Is DC one weak mechanism towards this, if people choose their projects according to their opinions of which direction science should develop?
[edit] How does Climateprediction.net do on these measures?
For Potential Impact, SETI will claim that discovering ET would be the most important discovery ever. Not too far behind that, climate change is the biggest threat facing civilization as we know it.
On Probability of that Impact where probabilities are brought into account, SETI does nowhere near as well. Climateprediction is also unlikely to solve climate change but it is highly likely to produce the world’s best probabilistic climate prediction. (It is producing 1000 times more models than any other ensemble.) With climate change being the biggest threat facing civilization as we know it, we need such information for politicians to make appropriate policy decisions.
I think Climateprediction.net does very well on Science. Unlike most (all?) DC projects, the Climateprediction project involves the scientists further analyzing the results of the work crunched. There are quite a lot of scientific papers that mention SETI. However most are about the way SETI has got so much computing power and how that could be used by other projects. There are also some papers about how Climateprediction is going to work but this is about to change towards papers about the results obtained.
It is probably pretty difficult to judge when a projects does well on Timing of the Science. Telling when a project is doing badly on Timing of the Science is easier. There have been times when SETI has handed out the same work units repeatedly rather than annoy their crunchers by saying there is no new work available. I think they have or are expecting more equipment so this may not apply any longer. I know there is more work that Climateprediction wants to do and then there is a sulfur cycle experiment to be added and then there is the hindcast experiment. There are lots of different stages and this provides interest and increases the chance of the project being held by lack of processing power.
On Use of Science, better knowledge of the probabilities of different size effects of climate change, can only be helpful for preparations for it. Scientific papers will be freely available in the usual way. The work is intended to become part of the contributions towards the next IPCC report which also makes its reports freely available.
On Safety First, I see no reason not to trust Climateprediction.net at least as much as any other project. MD5 checksums are provided for the BOINC download and I have not heard of any problems.
Personal Benefit - Sense of Community is going to be a personal choice. I think the Climateprediction boards are very interesting and are generally very good with there being very few flame wars.
Personal Benefit - Competition
Personal preferences dominate here.
Personal Benefit - Interest
With Climateprediction, you get a climate model and can watch how it evolves. Understanding what is going on can take some effort but many people find this rewarding. You do not have to get involved with this if you do not want to. This measure probably encourages people to move about from one project to another. This is probably the opposite of what Competition encourages.
Personal Benefit - Visually pleasing I have heard it suggested that short of DC that studies the body shape of 18-24yr old women/men, Climate Prediction is hard to beat. It is the only DC I know of that has an interactive visualization. You can spin the globe, view temp, pressure, rainfall, clouds and much more. There are even additional programs to view even more about your model if you want.
If I claimed that Climateprediction was the best project on all reasons, you wouldn’t believe me. There is bound to be an Achilles heel, and this may be it. Some people find their computer just doesn’t like the code and simply uploads lots of work units early. Backing up work to minimize any losses is often recommended. The code is complex (a million lines of FORTRAN) originally used on supercomputers and the nature of the task makes it very unforgiving of any errors. Consequently the code does need stable computers. Some overclocking of processors that appear to work fine on some projects may cause Climateprediction to produce errors. All this might sound bad but most people manage without serious problems and some people may see this as a challenge or a good way to test if their computer is really stable.
Political signal I think climate change is a really important issue and it would be good to have very large numbers of people crunching Climateprediction in order to send a message about the importance of this issue to politicians.
Democratizing science Again, I think climate change is a really important issue.
Most projects will have a few reasons against crunching for them. Climateprediction is not without any. As noted above the work units are quite difficult to complete. Also some people may see the large work unit as off-putting due to requiring a large commitment to complete them (4 weeks on a fast processor). However you do collect Credit at frequent trickle intervals and some people may take it as more of a challenge. One last advantage particular to the project – Work Units are always available.
