Recently, people have been worrying on Reddit and Steemit if BOINC (the distributed compute platform of Berkeley fame) can keep up with the 'demand' imposed on it by Gridcoin (the cryptocurrency rewarding scientific research). As a PhD student, I find the whole concern a little funny, but if you'd humour me I'd love to explain why this will never happen.
As research moves into the modern age, the computation requirements for the bleeding edge of most fields require exponentially more compute power. This will continue to be the case, meaning the compute requirements of progress will get bigger and bigger. If Gridcoin could ever truly provide a platform where any researcher can get any models run in a short timeframe, that would literally mark a new age in the history of humanity.
To put this in perspective, lets take a look at a minuscule corner of biomedical research. If you undergo open heart surgery, your body will be in a bit of a mess when you are recovering in the ICU. One of the complications is that you effectively become a temporary diabetic, unable to regulate your own blood sugar properly.
Blood sugar too high? You die.
Blood sugar too low? You also die.
For my biomedicine honours project, 4 of us spent a whole year developing a new insulin dosing regime for these patients. Developing a regime like this is very complex, as it takes a lot of inputs about the patient and needs to return customised dosing regimes for each individual using the same algorithm. You can probably image it is hard to tailor something to fit every person, but the process goes something like this:
We review past experiments and the literature, and combine that with our own knowledge and intuition to come up with a new regime.
We apply the regime to literally thousands of virtual patients. These virtual patients have all the characteristics of a real human, as they are individualised based off real people that have come through the local ICU in the past. These simulations take a LOT of computational power.
The results are reviewed, then we rinse and repeat ad infinitum. We were actually the 11th year of students to work on this specific project, and there are still students making improvements today.
Now, imagine if at the start of that study I was informed that I could get any amount of compute power I could possibly want, through the BOINC platform. We would have skipped step 1 altogether, and just made a BOINC project try every possible regime, every dosing frequency, every drug delivery method and combination thereof, and then tested the outcome on thousands of virtual patients. At the end of that set of simulations, we would have the perfect insulin dosing regime for patients after open heart surgery. The project would end then and there.
This is why I am frustrated to no end to see the amount of compute going to waste mining BTC and other POW coins. People want to improve the human condition, but don't want to help researchers get there.
There will never, ever be a point where BOINC cannot keep up with Gridcoin demand, until we have brute force cured every disease, cracked every encryption, mapped the entire known universe, etc. I'm dead serious.
If you are still not convinced, allow me to show you something else on a different track:
Do you see that tiny speck in the right-hand sunbeam? That is the Earth. This photo was taken by Voyager 1 on February 14, 1990, at a record distance of 6 billion kilometers. We are specks of dust on a speck of dust. Even ignoring all other forms of science that need compute power, we need an unfathomable amount more computing power to even begin to understand the universe we live in - something that will be vital for the longterm survival of humanity.
Please, do the research community a favour, and use your compute power for something that actually makes a difference.
Content Credit:
Pale Blue Dot, NASA
