Originally published at: https://boingboing.net/2020/07/27/this-simple-explainer-tackles.html
…
Thoughtly’s video takes extremely long to (AFAICT - I skipped around because it is so painfully long) to get to basically nothing except reinforce misconceptions. The “quantum computing uses probabilistic behavior to be massively parallel” which is at best naive and as presented terribly misleading. The MIT technology review article is blessedly shorter, but doesn’t get it much better, although at least it doesn’t go into excruciatingly painful and misleading detail about how “300 qubits can processes 2^300 variables in parallel”
If you want a short but not wrong description of quantum computing this is much better than either of those options:
13 Likes
Um, I don’t get it…
1 Like
Yes, exactly this.
A few years ago I was able to sit down with the head of IBM’s quantum computing research for about 2 hours to chat about his work, and I have to say, getting much beyond the level of that SMBC comic would be really, really hard without already having at least a few semesters of actually-well-taught quantum mechanics classes, as well as basic linear algebra, multivariable calculus, some computer programming skills, probability, statistics, information theory, and algorithms.
5 Likes
The most important thing, which is conveyed in the SMBC comic is that amplitudes are different than probabilities. Amplitudes can be positive, negative, or complex. When you measure a qubit, there is a rule for converting amplitudes to probability, but when you are doing quantum gates those amplitudes add directly. That means that a positive amplitude can cancel with a negative amplitude and cause destructive interference.
So the important thing about any useful quantum algorithm is not so much that it creates a massive superposition of many possible states. The important thing is that it manages to have a large fraction of the ones you aren’t interested in destructively interfere before you get to the measurement stage. IMO any description of quantum computing that leaves this out doesn’t belong in 2020 even for a layperson audience.
You can actually construct a machine that is restricted to creating quantum states with only positive amplitudes (there is a bit more technical detail to this, but that is the basic idea). This type of machine implements very close to what we think of as probabilistic computing and in fact is proven that it does not have any computational advantage over a classical computer with randomized algorithms. You can simulate these systems efficiently with a technique called Quantum Monte Carlo which basically does probabilistic evaluation of the quantum state. The DWave quantum annealer is an example of this:for a long time their devices were only able to generate positive amplitude superposition. This is one of several reasons why almost all quantum computing experts outside of DWave were convinced that their machines weren’t able to do anything dramatically better than classical computers, even in principle. They have more recently being to work on better designs that try to add capability for more powerful operations, and while there are still some fundamental issues in their approach it is a long overdue step in the right direction.
3 Likes
I haven’t watched the video yet, so I won’t comment on that, but I love pictures of quantum computers. They seem kind of steampunk.
2 Likes
Well I still don’t get it, but I get it lots better from this comic than from the video! Thanks!
2 Likes
This comic is the BEST THING IN THE WHOLE UNIVERSE!
Okay, it’s not.
It’s both, simultaneously… (ducking for cover)
Great explanation, though!
Also I want this panel on a tee shirt - 
s
(repeating the link to the original by Scott Aaronson and Zach Weinersmith)
5 Likes
Because it does take a while to get into the subject,
5 mins in and still on Moore’s Law?
Took too much brainpower just locking onto his accent.
While the technical use cases for QC are interesting, it is the scientific realism/ philosophical implications that interest me more. “Shut up and calculate” is all very well on a theoretic basis, but when you actually start doing things with QC it becomes a lot more untenable to insist that the UWF/Everett manyworlds etc is “just a mathematical device”. You can’t do work with mere mathematics. If these calculations are happening, they have to be happening somewhere which requires an ontology. Explaining where then becomes a valid question in terms of Copenhagen vs. Everett, and it gets a lot harder to reply “don’t ask”.
Is “polynominal” a common variant of “polynomial”? I haven’t heard that before, and my go-to search engine suggests it’s just a misspelling of the latter.
Moore’s law is misstated. Arguably, it’s just oversimplified.
The video uses lots of stock footage to illustrate things, which isn’t necessarily a bad thing, but it’s often not made clear when these clips are meant to be suggestive rather than literal.
For example, that was a generic gravestone shot where it talks of Babbage’s death. Babbage’s actual grave available, which has an entirely different style of monument. And most of the pictures of his completed difference engine are of modern recreations of his design for a larger difference engine–one that proved too difficult to build in his time. The engine that he did complete was much smaller.
Individually, none of these nits is serious. But the totality of them cause me to wonder whether the presenter’s knowledge of the subject goes much deeper than having read a couple Wikipedia pages. I want my explainers to come from experts who can distill the topic to the salient points that I, as an outsider, will find interesting and useful.
1 Like
This topic was automatically closed after 5 days. New replies are no longer allowed.
