|
I just finished reading Atlas Shrugged, by Ayn Rand. I've tried reading it before (maybe 15 years ago?) but I don't think I got through it. It didn't make much of an impression on me at that time, at any rate. This time was quite different. I think my understanding of the world has changed significantly and some parts of the book really resonated strongly with me.
It's a bit of a mixed bag, I guess. The bad part is that the world seems to be becoming more like what Rand described in the book. The good part is that the book was written a long time ago and Rand probably also felt the world was like that then, so maybe it's not actually trending worse but just feels that way.
In terms of writing style I found the book rather verbose, and I did skip over some of the longer philosophical speeches once I got the gist of it, But still definitely worth reading.
[ 0 Comments... ]
(No this is not a post about commercialization of AI)
It occurred to me today that there is an interesting parallel between AI and capitalism. They're both tools that can be used to accomplish things very effectively. And they both have "alignment" issues.
Capitalism provides incentives and a free market, and that can result in extremely efficient action because people are trying to maximize profit. However, left unchecked, this can produce all kinds of negative externalities which are not really what we want. This is the alignment issue - we need to be diligent in "aligning" capitalism so that it incentivizes and produces outcomes that are actually what we want. Primarily governments do this with the use of taxes and subsidies - make bad behaviour (produces negative externalities) more expensive, and make good behaviour (produces positive externalities) less expensive. In my view, this is the primary function of a government (at least in a capitalist society). However, I think most people would agree we haven't yet perfected this part of things, and we still suffer from all kinds of negative externalities.
AI/AGI is very similar - it can produce extremely efficient action, but suffers from an alignment problem in that what it does may not be exactly what we want it to do. Lots has already been written about the AI alignment problem.
My main point here is that we, as a society, haven't even figured out how to align capitalism reliably. Which makes me a bit worried about how we're going to handle AGI. We should probably think hard about why we haven't been able to align capitalism perfectly (e.g. maybe because not all subsets of people want the same thing?) and see if those reasons carry over to AGI as well.
[ 2 Comments... ]
Ever since I read The Information I've been noodling on the idea of abstraction. By this I'm specifically referring to the software engineering concept, where you have a bunch of complexity that you hide behind a simplified interface at different layer of abstraction, usually as a means of managing complexity.
That's a lot of words but it's something we do and use literally all the time. Here's a simple everyday example: say there's an apple in a cupboard and you want it but can't get it yourself, so you ask somebody else: "Would you please bring me the apple in the cupboard". This is an instruction at a pretty high level of abstraction.
You could accomplish the same goal using lower-level instructions, such as "Would you please walk over to the cupboard, open it, take out the apple, close the cupboard, walk over to me, and put the apple in my hand." Whereas the high level instruction was more goal-oriented, we can think of this as more action-oriented.
And you could imagine going into even lower-level abstractions by describing how to e.g. contract specific muscles or activate specific joints in your body to accomplish those actions. This might be considered a mechanism-oriented level of abstraction.
A couple of interesting notes on this.
First, the concept of "high" and "low" level abstractions is relative. In the example above, an even higher level abstraction would be, for example, "I want the apple to be in my hand." This might be considered a declarative- or state-oriented level of abstraction. That being said, this is still a very wishy-washy description of a level of abstraction, and I would love to see a more rigorous/numerical definition of it.
Second, given the nature of the problem, some abstractions are more appropriate than others. I would argue that solving problems at the correct level of abstraction is the best way to ensure the problems "stay solved" in the sense that the solutions don't need to be revisited or tweaked over time. I have found this time and again in the domain of software engineering and believe it applies in other domains as well. I'm specifically thinking of the way laws are formulated in response to problems, but often have unintended consequences because they are targeting the wrong level of abstraction and don't generalize well to new situations.
Finally, I believe that the concept of abstraction is severely under-emphasized and under-studied. Finding a way to measure the concept of abstraction numerically, the way Shannon did with the concept of "information" will unlock whole new fields of science and research. Bring able to quickly find the right level of abstraction at which to solve societal problems means that we can save tremendous amounts of time and pain, so the potential upside is huge.
[ 0 Comments... ]
Over the past few years we've been reducing our direct use of fossil fuels in our household. This involved switching out three fuel-powered machines for electric equivalents: our car, water heater, and home heating.
The car was the first one we did (in late 2019), and was the one with the largest up-front cost. But given the supply chain shortages and rising gas prices, I'm glad we got that one done first. I haven't done a detailed financial analysis but I know that our insurance went up (simply because the base price of the EV was a lot higher than the ICE we had before) and all other costs went down. Overall I think financially it's a slight win in terms of operating costs, but the payback period for the up-front cost will be pretty long. However the pleasure of driving an EV counts for a lot! It just feels nice, even now 2.5 years later.
The water heater was the next one to go (in mid-2020). Previously we were renting a gas water heater; the rental company was not the best and I was eager to ditch them. After doing the math I concluded owning was strictly better than renting, so we purchased an electric water heater and had it professionally installed. The payback period on this was quite short, and we're already net positive two years later. Even if we have to replace it every 6 years (the warranty lifetime) it's still a win over renting. Plus the electric heater is new and doesn't look like it's going to explode if I look at it wrong (the old gas heater was quite old). Functionally it's equivalent or better - water is hot.
And finally, last week we replaced our nearing-end-of-life gas furnace and AC with an electric heat pump (plus a new gas furnace as an auxiliary heat source). It's only been a week but so far I'm pretty happy with the heat pump. Based on the rated capacity, I expect the backup gas furnace to only kick in a few days a year. The real test will come next winter though, so I'm going to withhold final judgement until then.
We still secondary fossil-fuel-based things like a propane BBQ and gas fireplace, but they're very infrequently used. It's quite satisfying to have all our primary energy needs met by the electrical grid. In Ontario at least most of our electricity comes from non-carbon-emitting sources, so that's a nice bonus.
I keep wondering what life would be like if things like cellphones and laptops ran on fossil fuels. Imagine having to open up a little port and pour in a thimble of oil into your phone every night instead of just plugging it in. I'm glad we live in the future.
[ 0 Comments... ]
Imagine a 3D printer that sucked out carbon dioxide from the air, extracted the carbon atoms, and generated graphene/nanotubes in your desired configuration. Presumably with sufficient energy and some sort of catalyst this should be possible. This would be amazing! Not only does it create products with the properties of graphene (amazing strength, etc.), it fights climate change while doing so! And it's a 3D printer that doesn't require you to procure physical "raw materials" since it would suck it from the air.
A 100g print would require 100g of carbon, or ~367g of CO2. At the current atmospheric CO2 levels of ~420ppm, that requires around 873kg of air, or ~713 cubic meters of air. Assuming a 2.5m ceiling height, that's about the amount of air in a ~285 square metre house (~3068 square feet). That's a pretty big house, but a not-unreasonable order of magnitude. Hook up some giant fans and you're all set!
One day we'll probably discover that trees do exactly this.
[ 0 Comments... ]
I recently read "Termination Shock" by Neal Stephenson. Good book for sure, although I prefer some of his other works.
There was a bit in the book that involved deepfake videos which I thought was interesting. Making a deepfake video of some authority figure (government, etc.) and spreading it as misinformation seems like a pretty bad thing, but it's also something that can be defended against. All you have to do is have authentic videos include a public-key signature that can't be faked. So e.g. the US government could have a published public key, and any official communications would be signed by the corresponding private key. Media distribution platforms could then verify the signature to authenticate that it hasn't been faked or modified.
In fact, this would be good even without the threat of deepfakes, because the signature would apply to the entire video rather than a soundbite/clip of it, and so if you extracted and shared a clip, the clip would not be verifiable. This is good because extracting a clip often loses important context and is not representative of the "whole truth". Sharing this kind of out-of-context information is in some cases equivalent to sharing blatant misinformation. Organizations that create the original video could still create shorter "approved" clips and sign them for sharing if they desire.
Same applies to any kind of digital document, really. That's why people do it for email!
[ 0 Comments... ]
Over time we've accumulated a fair number of black plastic takeout containers. The problem is there's a bunch of different sizes, and finding the matching lids can be quite annoying when they're all just stacked in a drawer. Today after much frustration I (with my trusty 3-year-old assistant) finally got around to sorting the containers to match up the lids and discarding all the pieces without matching parts.
The next step to prevent a re-occurence was to label them. Labeling the lids was easy enough with a permanent marker, since the lids are clear. However the containers are black plastic and none of the pens or markers I tried made any easily-visible mark. In the end I used a knife to make notches in the plastic, with the number of notches matching the number on the corresponding lid. This works pretty well; once I have a container I can quickly count the notches and find the matching lid.
I only wish I had thought to stack the containers by size before numbering them. As-is, lookup is O(n); I could have made it O(log n) using binary search if the lids were stacked in numerical order. If I stack numerically now, they don't stack well because larger ones get put inside smaller ones. But the constant factor is much smaller than it used to be, and that's the real win here. Adding a new container is now O(n) (vs O(1) previously) as I have to compare it to previous containers to determine which number it needs to get, but that should be a relatively infrequent operation.
[ 0 Comments... ]
An author of popular free software packages intentionally inserted infinite loops into his code to break downstream users, as a form of protest. It's not the first time this sort of thing has happened, and it won't be the last. In this particular case I would say that both primary parties are in the wrong.
To the author: if you make your software freely available, obviously people are going to use it without paying for it. Why would you expect anything else?
To the users: if you use some random piece of software in your code without verifying it and just keep blindly updating it, obviously stuff like this is going to happen. Why would you expect anything else?
To me it seems like there's no strongly defined interface here, on either side. So you have to be prepared to accept any kind of behavior from the other party. What's surprising is that this doesn't happen more frequently.
What I would like to see is some mechanism for a formal "pay what you can" style of money transfer. As an author of various open source packages, I do it mostly for personal pleasure, but I also wouldn't be opposed to getting money from those that profit from the work. This is not my primary concern, but if it were easy to set up, why not? On the flip side, as a professional software developer working for a profitable company, I would love to be able to support the (large amounts of) open source work we rely on.
Approaches like Patreon are fine for "artisanal" stuff but they don't scale. What I want to be able to do is take a budget of any amount, say $1m and distribute it programmatically across the free software projects that we use. Ideally with most of it actually going to the project as opposed to getting sucked into transaction fees or middlemen.
So here's a strawman proposal: open source projects that are accepting donations should include a DONATIONS.txt file in their source tree, similar to README.txt, that provides a machine-readable money transfer address. These files can be scraped, collated, and funded by any consumer of the project.
The hard part, of course, is the "machine-readable transfer address" part. Obviously cryptocurrency seems like a good fit for this problem. It's trivial to create a wallet and let it accumulate donations, and only go through the hassle of extracting the value in fiat if there's enough to be worthwhile. And it works at internet scale, that is to say, regardless of what geopolitical area of the world the parties are in.
For people opposed to cryptocurrencies: sure, there are good reasons to be sceptical. But I challenge you offer a viable alternative instead of shooting down this proposal merely on the grounds that it involves cryptocurrencies.
[ 1 Comment... ]
I've been thinking lately about the revolutionary new technology transitions that humanity is going through. There's a few that I think will have a huge impact over the course of the next few decades:
- artificial intelligence, which is already well under way
- quantum computing, which is still early days, but will give humanity unprecedented ability to effectively engineer everything from materials to genomes
- space exploration/colonization, which will be a driver for creating all sorts of new products that will benefit everybody
There's also a few shorter-term things like cryptocurrency/web3, 3D printing, and global internet coverage which are similar in nature but smaller in their impact.
What's interesting to me is that while all of these things are in development concurrently, I'm not sure humanity has enough bandwidth to digest all of these revolutions concurrently. In isolation, each piece of tech can be developed just fine. However, there's going to be a lot of work needed to integrate it with all existing tech and with each other. The "each other" part of it is particularly expensive because for n new technologies there are O(2^n) combinations and exploring each of those possibilities requires knowledge of one or more of these new technologies (where, by definition, the knowledge is not widespread). I'm not sure we have enough people to make it happen smoothly.
It will still happen, just less smoothly than we might like. By that I mean gaps will be exposed at the intersection of technologies that can and will be abused by bad actors. Fortunes will be made and lost to these gaps.
If this all sounds very abstract here are some more concrete (although rudimentary) example questions:
- Who will update software to move off cryptosystems like RSA that can be broken by quantum computing? What will get left unpatched and hacked?
- As products get lighter (mass efficiency being a key factor for space colonization) delivery via drone becomes feasible for a much larger set of products. Who will have the drone delivery capability to take advantage of it?
- How do blockchain consensus protocols deal with the scenario where a bunch of nodes are on another planet with multiple minutes of latency in communications? Is a single unified system still feasible?
These questions are pretty basic and I don't think I have enough specialized knowledge to ask better ones. But I know they're out there, and they're important to think about.
Extrapolating on this line of thinking, there's probably some ideal ratio of scientific/research activity to engineering activity such that the rate is creation of new technologies doesn't overwhelm the ability to integrate those new technologies into the world. Over time, this ratio would increase in the direction of requiring more engineering activity per unit of scientific/research activity simply because each new technology can be integrated with the increasing number of technologies that came before.
[ 0 Comments... ]
Over the course of 2022, pick a book from every major Dewey Decimal category and read it. This probably works best if your local library uses Dewey classification rather than Library of Congress or other classification systems, and if COVID isn't preventing you from actually browsing the library in person.
[ 0 Comments... ]
|