8 min read

Friday Fodder (7/24)

Why you should ignore pundits, Google breaks its brain (or how to slowly ruin a good business), Australia's Workplace Gender Equality Agency released data that are about as useful as tits on a bull, and is Joe Biden too old?
Friday Fodder (7/24)
Artwork by DALL·E 3

Please note that there will be no Monday newsletter over the next fortnight due to various public holidays around Australia. Normal service will resume on Tuesday.

1. Pundits should always be ignored

Warren Buffett's annual letter to shareholders was published last weekend, and as we've come to expect it contained a few gems along with some words of caution. For long-term investors like his sister, Buffett says:

"She follows business news – reading four newspapers daily – but doesn't consider herself an economic expert. She is sensible – very sensible – instinctively knowing that pundits should always be ignored. After all, if she could reliably predict tomorrow's winners, would she freely share her valuable insights and thereby increase competitive buying? That would be like finding gold and then handing a map to the neighbors showing its location."

The same could be said about people such as Rich Dad, Poor Dad author Robert Kiyosaki, who wrote a book that "was pretty short on specifics of how to actually do this [make money in rental real estate], beyond recommending expensive courses offered by Mr. Kiyosaki". Beware taking financial advice from people who make a living by selling that advice!

But back to Buffett, who is now sitting on a lot of cash (a whopping $167.6 billion) partially because his firm, Berkshire, is now so big that he's finding it hard to buy anything that will "move the needle":

"Berkshire now occupies nearly 6% of the universe in which it operates. Doubling our huge base is simply not possible within, say, a five-year period, particularly because we are highly averse to issuing shares (an act that immediately juices net worth).

There remain only a handful of companies in this country capable of truly moving the needle at Berkshire, and they have been endlessly picked over by us and by others. Some we can value; some we can't. And, if we can, they have to be attractively priced. Outside the U.S., there are essentially no candidates that are meaningful options for capital deployment at Berkshire. All in all, we have no possibility of eye-popping performance."

That is until the next crisis, which "will cause stocks and bonds of some large and fundamentally good businesses to be strikingly mispriced... [and] Berkshire's ability to immediately respond to market seizures with both huge sums and
certainty of performance may offer us an occasional large-scale opportunity".

It's tough to achieve high rates of growth when you get to be as big as Berkshire, so trying to mirror its trades probably isn't a winning bet. But you can be sure Buffett – or his successor – will be ready to deploy his huge stock of capital the next time there's a crisis, as he was in 2008/09.

2. Google breaks its brain

After an upgrade to Google's Gemini AI (formerly Bard) – version 1.5 now comes with a huge 1 million token context window (~700,000 words) – people started to notice that, well... they weren't exactly getting the results they were expecting:

Internet sleuths were quick to test Gemini in other areas, soon finding that it would also refuse to make positive comparisons or create pictures of white people in virtually any context. But curiously, it had no problem drawing up America's founding fathers, Vikings, and literal Nazis as... non-white. Even Google's founders were depicted as two Asian men.

Google's very expensive, very powerful Gemini chatbot had started to resemble Robocop 2, where "all sorts of 'socially positive' silliness" was programmed into him, breaking his brain and rendering him useless.

Many of the issues above have now been fixed and Google has delivered an apology. But it does raise the question about AI going forward: how useful is a model that can't avoid imparting the political bias imposed by its aligners into its answers? Do we want the truth, or Google's version of the truth?

For consumers of AI, the good news is that no company has managed to build themselves a proverbial moat; there is already an ocean of large language models out there, and barring regulation that 'locks-in' a particular outcome, competition should ensure we avoid a Google Gemini becoming the only option.

The fact Google had to apologise and make changes so quickly indicates that it's worried about losing ground in this space, although the whole saga looks to be another sign that management is running amok at Google, prioritising everything except making a good product.


Google's lucky it still dominates the search market, which allows its management to get away with a bit of discrimination without having its profits competed away. It got to that position by creating a fast, efficient and relatively neutral search engine that people love so much it even became part of our everyday language: "Google it". But if Google thinks it can be a dominant force in AI by using it to force an ideological agenda down our throats, it had better think again.

3. Tits on a bull

Australia's Workplace Gender Equality Agency (WGEA) published a dashboard containing a trove of private sector employer gender pay data on Tuesday, with the headline figure showing a gender pay gap of 21.7%. It's an impressive amount of data and you can even break it down by company; for example, here's a snippet of the pay gap and gender composition of Australia's largest corporate, BHP:

As far as dashboards go, I give it a 7/10; not quite as modern, fast or user-friendly as some of the others I've seen but it gets the job done. By government standards it's probably closer to 10/10.

But for policy making? About as useful as tits on a bull.

The problem with data such as these is they are displayed with the authority of a government agency; they are "the facts". But without providing all-important context they can mislead certain individuals, including the commissioning editor for news.com.au, into making claims such as "[t]his latest report proves (again) that the gender pay gap is real and women are being discriminated against".

But the report does no such thing. For example, not once does the WGEA mention hours worked, when those hours are worked, or whether the participants had taken career breaks for, say, child-rearing. Women also tend to self-select into professions and roles within professions that offer more flexible schedules, which can come at the cost of total pay.

There really is no excuse to not, at a minimum, include footnotes acknowledging the complexities of these data. It's not hard to find the evidence: Harvard's Claudia Goldin literally won the economics Nobel Prize last year for her work on this topic.

It's times like these when I wish we had more John Cowperthwaite's in government, Hong Kong's former Financial Secretary who refused requests from London to collect some data "because he was convinced that once the data was published there would be pressure to use them for government intervention in the economy".

If you can't do it properly, sometimes it's best to not do it at all.

4. Is Biden too old?

US President Joe Biden is 81 years old. By the time of the next US election, he'll be about to turn 82. Biden is planning to run for another term, which if successful would make him almost 86 by the time he retires. Given his age, the statistical probability of death during that time is about 8.8% each year, which over the next five years would equate to around a ~37% chance (1-(0.912^5)) .

If you're a vote that's hard to ignore, especially as Biden's vice President, Kamala Harris, has consistently underperformed her recent predecessors in terms of electoral favourability.

So, is Biden too old for voters to trust him with another term? Prominent Democrats such as Matt Yglesias don't think so, at least if his list of 21 thoughts on Biden's age is anything to go by:

"We're dealing here with a world of compounding uncertainties. Where Ezra and I agree is that Biden is currently on track to lose the election. And I agree with Ezra that if Biden loses, which he probably will, age-related recriminations will almost certainly loom large in the aftermath. But Ezra has what I think is an unreasonably high level of confidence in the proposition that Biden stepping aside would lead to a stronger nominee, because I believe he is overrating the genuine importance of age relative to issue positioning in explaining the pattern of public opinion. I have a very high degree of confidence that Biden could improve his standing by positioning himself more moderately, and a very low degree of confidence that Biden stepping aside would generate a better-positioned nominee. And I very, very strongly believe that Kamala Harris should seek to aggressively moderate her image, but my understanding of the situation is that she sincerely does not want to play the role of identity politics human shield for the white male moderates of the Democratic Party.
My advice to Joe Biden is that he should appear more in the media, and that instead of his team being afraid of 'gaffes' and obsessed with party unity, they should embrace the more moderate framings that come more naturally to him and that reflect his long career espousing pre-2016 Democratic Party ideas."

FWIW, Joe Biden is slightly trailing Trump in the polls but is behind by a long way in the betting odds. Unlike Yglesias, I'm not so sure Biden can get out in the media more and avoid 'gaffes', which have only increased towards the end of his Presidency; perhaps hiding from the media is his best hope. But then I'm no political strategist.

If Biden does lose the election as most are predicting, then the most important question for Australia is what might a Trump 2.0 Presidency look like? Trump will be older than Biden was when he was elected to his first term but doesn't appear to be as prone to gaffes, although as Reagan showed, mental decline can come on quickly. But Trump still has more than a few screws loose: from what he has publicly said so far, we'll probably see a repeat of the last time he was in office, including a slew of protectionist tariffs and large fiscal stimulus (tax cuts with no offsetting cuts in spending).

Without the threat of another election restraining him (not that Trump could ever be restrained), the temptation will be for him to go even bigger on all fronts. I don't know how that will end except that if "Tariff Man" launches a fresh trade war then we'll all be worse off in the long run.

5. And if you missed it, from Aussienomics

Bowen's taking us for a ride – Bowen's proposed vehicle emission rules rely on dodgy assumptions, restrict choice and raise prices while overstating the savings. On the list of things to do about carbon emissions and the cost of living, they are one of the most costly, inefficient options available.

A nuclear future for Australia? – Ditching coal is inevitable for net-zero, but is nuclear the answer? Maybe, maybe not. But what we need is an honest discussion about our energy future.