You were right
You were right when you said “bitcoin will never overtake the USD and can’t cope with large transaction volume”.
You were right when you told everyone not to invest in it because it will ‘crash’ and ‘it’s a bubble’ and stay safe and buy Westpac shares.
It hasn’t overtaken the USD, it did crash on multiple occasions and its base layer transaction processing capability is unchanged. Yet, over the same time period of your rightness, it has still increased 1000x, 50x, 5x or similar and likely outperformed any other investment that you substituted for it. Why?
The trend of rejecting anything new is the only thing that isn’t new. Personally, I made the mistake with social media. I loathe it and so rejected it as an investment proposition because I didn’t like the fact that it invaded privacy, stole data and very obviously was the monetisation of people’s personal lives. I was completely right, and missed one of the biggest investing trends of the last decade.
Now the same thing is true with AI. The mega-clever crowd has arrived to either tell us it will kill humanity, or that it’s a nothing-burger, the more radical the take on either end the better.
Inevitably, Chomsky has weighed in to tell us that GPT is nowhere near as clever as he is. He lays out that it is “just” a pattern recognition algorithm, it simply looks for patterns in massive data sets and then guesses at an answer that might suit. He contrasts that to the human brain which looks for “explanations”. No explanatory power = Not clever.
His description of the process of AI is 100% correct. He will also be massively wrong because no explanatory power does not mean not useful. The fact of the matter is that GPT and many of the other AI tools are incredibly useful. It’s the fastest growing platform of all time and not for no reason. Does it really matter that the tool is guessing with high accuracy. Which human isn’t?
Here’s an example. I received access this week to the plug-ins on GPT4, which include Wolfram Mathematica, a very powerful maths program. I googled a maths question1.
Several things have happened here. ChatGPT has correctly synthesised the question. It has then passed that to another, more powerful, mathematics program and communicated with it in a language I am not required to learn any longer (yes, Wolfram has its own language) and returned the solution. That’s kind of impressive but perhaps not that special.
It does get better though because it can go on and explain how it got the answer.
Let’s assume further that we don’t fully understand the notation. We can ask it a fairly dumb question that might be a bit embarrassing in an open forum. Without judgement, it explains.
I can see how this is incredibly useful to the learning process because now we can ask questions and not be embarrassed to ask them. One of the major problems of classrooms is that there is always a scenario where you are too embarrassed to ask because everyone else seems to immediately understand. You nod and move on and learn nothing. You can continue to dumb down your questions to the base layer and it will continue to explain, without losing patience.
My point here is that people who are sneering on the sidelines about AI probably haven’t even scratched the surface of using it. It just sounds erudite to knock it because you are making the claim that you are cleverer than the machine. We get it, you are, but I find very little value in the people who are saying so.
The thing triggering everyone seems to be the ‘intelligence’ aspect. Whether or not something is intelligent seems to be rather important. If that is so, I present on the left a dolphin brain and on the right, a human brain. The dolphin brain is twice as large and their intelligence is well-known, yet nobody has yet appointed a dolphin as their CEO. Why is that? It would be great wouldn’t it? Flipper will now deliver our five year strategy.
We should put aside the obsession with intelligence and embrace our discoveries. The formula underlying all AI is simple enough it could be taught in high school. It is unbelievably short and effective and calculates the relative weights of inputs and their likely relevance. It is how AI guesses, as Chomsky puts it, the correct answer.
It was laid out in a paper by researchers at Google in 2017 and its simplicity alone should make you wonder.
My guess is that the formula will become famous. My guess is that the uber-clever naysayers will be right and that this iteration of AI does not surpass human intelligence and is a mere step along the journey. GPT will disappear and just be integrated behind the scenes and you won’t notice. They will regularly remind us of their rightness.
My guess is also that they are very bad investors.
1 I gave it another question too which it got wrong and then corrected. Full details here if you are interested.
2 The Softmax function is an invention in itself. It converts the weights into probabilities so that they sum to 1. In machine learning, the term “softmax” is credited to British academic John S. Bridle who first used it in machine learning in 1989. His original paper is here. It goes back further than that though, the technique itself was discovered by Boltzmann in 1868.
Weights and measures
Chomsky-style then, in the continued search for good explanations I found this. The internet was musing last week on “what does god think of bitcoin?” Responses varied, but included this.
A false balance is an abomination to the LORD, but a just weight is his delight.” – Proverbs 11:1
“A just weight”. A good explanation indeed of why fiat currency always fails in the end. Why would anyone work for $1,000 only to get home and realise it is $998 because 2% inflation is “good for me”. Wait until they realise that the 2% is actually 15% and their $1,000 is really $850.
A just weight is one that cannot be affected by someone else to their own or someone else’s advantage.
In the digital age, a just weight needs to involve computation, it needs to involve energy conversion. It must be invisible and portable and yet inextricably linked to the physical world through its link to energy. If I gave you those parameters I think it highly unlikely that you or anyone else would be able to solve that problem.
Someone did though and now we have a thermodynamically pure measure. It is the thing putting the wrecking ball through the many lies of modern finance. Maybe that’s good, maybe it’s bad; but the truth, now that it’s out, cannot be contained.
Permanent for the near future
Speaking of 2%. Bulgy eyed Federal Reserve psychopath Neil Kashkeri was back on the wires this week.
Kashkeri is the man behind one of the greatest thefts of time. The Troubled Asset Relief Program (TARP). Without delving too deeply into the story about him and his Goldman Alumni chums enriching themselves it is sufficient to say that his reward for his service to banking was a seat on the Federal Reserve Board. His CV, even the edited Wikipedia version, does not justify the position. It is strange, and basically tells a story of someone who has just blagged his way to the ‘top’. His media performances imply a man way out of his depth.
He has presided over a number of disastrous interviews during his tenure including his mind blowingly stupid “infinite cash at the Federal Reserve” performance. With my very low regard for him now firmly established let’s look at this latest contribution.
Is there a better euphemism for ‘not permanent’? You can watch him on Bloomberg blurting it out and then backtracking fantastically. He mumbles all sorts of stuff about ‘elected leaders’, ‘Washington’, etc.
There is no doubt this is a live discussion at the Fed about how they get the debt stack under control. Population dynamics mean they need an inflation rate somewhere around 4% or higher so debt doesn’t spiral. The best way to do it is to pretend, as they do now, that inflation is much lower than it actually is but that can’t work forever and a subtle change of strategy might be needed down the line. It won’t be obvious, it will be much closer to an ‘average target over a period of time’ as his rather more intelligent colleague Lael Brainard once laid out.
History will not be kind to Neil Kashkeri. To coin an American phrase, I would fire his ass so hard he’d be job hunting in another time zone.
The slide is from a speech by Stanley Druckeniller. It explores further a point discussed last week about the likely path of US interest payments. They recently exceeded defence spending, but he points out that by 2050, interest payments will exceed the total of all other outlay with 10 year rates at 5%. His assessment is actually quite conservative because it assumes that there is no increase in Federal spending beyond today and that the Federal Government is not called upon to bail out bankrupt states (I suspect they will be).
His prescription just seems too sensible for me, the United States should spend less. It’s the kind of disgusting talk nobody wants to hear; he must be mentally ill.
Is there the political will in America to cut entitlements among older cohorts? I don’t think so because they happen to be the largest cohort and you cannot win elections without them.
Will there be an outbreak of goodwill amongst the boomers to accept slightly less? No chance. Young people are lazy no-hopers don’t you know.
We know how this plays out. Buy something scarce, back the advance of technology, preferably combine the two. Then sit back and enjoy the show.
The European Union is having a bit of a rough time with regulation. They do love to regulate as we know and they love their petty rules which are good for “all of us”. Occasionally it can yield good outcomes like unpasteurized cheese, but mostly, it doesn’t work.
As time rolls on, new jurisdictions emerge that are more welcoming to new technology. Money shifts around and follows where progress is greatest. Sadly, that is no longer Europe. Google has had a terrible time battling the EU. They paid a $2 billion fine last year for antitrust breaches. The European Commission was delighted and couldn’t trumpet their triumph loud enough.
As a result, Google has been rather wary to launch its AI products in Euroland. 180 countries have it, but none of them are in the European Union. It’s not a meaningful disadvantage but it’s probably incredibly annoying for people that would love to use it. On the off chance that this technology is key to productivity gains it will leave Europe lagging behind, again. Google had this to say:
“Bard will soon be able to support the 40 top languages, and while we haven’t finalized the timeline for expansion plans, we will roll it out gradually and responsibly, and continue to be a helpful and engaged partner to regulators as we navigate these new technologies together.”
A helpful and engaged partner to regulators. That means, not yEU (ho ho).
The same story was true of digital currency. There was hesitation, outright attack and then heavy handed legislation. There is absolutely nobody in their right mind starting an AI or digital asset business in the European Union, why would you?
This chart is perhaps unrelated, but there is no migration of millionaires in any significant number to Europe. Despite its ‘sophistication’ and ‘unified currency’, people only go there on holiday and strictly for one month per year. Switzerland and Portugal garner a few thanks to their beneficial tax regimes but other than that not much happens. The UK wore the continued cost of its EU departure too, 1500 in net losses. Autocratic regimes fare even worse of course.
The country that attracted the most millionaires in 2022 was Australia (I’m not counting the UAE because net migration there is dominated by returning emigrants).
80,000 millionaires have moved to Australia in the last two decades. Is it surprising? Probably not. People from Western cultures likely find Australia a comfortable home and a nice gateway to the East. It’s 8 hours to Singapore from Sydney, not exactly next door but not the end of the world either. 30% of the population are immigrants born overseas so being “foreign” isn’t so foreign here.
Australia has its problems, which will grow with immigration, but I’d back it over the EU any day of the week. We have Bard too, which is nice.