roundball100

Do we have actual evidence that AI generates profits? Or do you mean that there is revenue (that is, paying customers beyond all the free services to the general public) but not enough to offset expenses incurred by the growth of the AI industry at this point?

or -> our

Does this mean that as FFH investors, or reward will be in the afterlife also? Just kidding.

What you did not mention in your list of personal circumstances is an individual's level of comfort with risk (here I use the risk to mean, for FFH, volatility in share price rather than permanent loss of capital). If for personal reasons you feel that you may have to sell in the near future for any reason, then you probably should not be invested in any stocks. Which means if the markets go up, you miss out on return, but if the market falls, you come out ahead. No one on the board can predict the future. But it sounds like you will sleep better with one of these choices than the other.

A slight variation on the mental model of a stream of water flowing out of the forest (carrying harvested logs) is a stream of logging trucks emerging from the forest on a logging road. The volume of harvested logs varies each year based on logging conditions and what types of trees are mature and/or sufficiently valued by that year's market to motivate cutting them down. While floating logs out by water may be more environmentally friendly or cost-efficient than shipping them by truck, trucking might become more cost-efficient if the trucks are switch to EVs (also reducing deaths arising from the notoriously dangerous job of log-rolling).

Please let us know when you next plan to step away from your computer again ...

How do I get access to such reliable charts? And all this time, I thought that stocks were unpredictable!!

Running open source models locally also makes a whole lot more sense from a privacy perspective (avoiding leaking the crown jewels, proprietary and private data, to an untrusted and often outsourced third party) cloud service.

Do we have to sign up for all 3 as a lot?

If we are trying to compare FFH to other companies, and FFH is able to average n% investment return reliably with 3:1 leverage (without undue risks), then should you not give them credit for how much this return increases their book value per year?

Does the 3:1 leverage give us 3 times that?

The one I remember about him (if I have it right): "I wouldn't join any club willing to take me on as a member ..."

And as you have suggested: massive disconnect = (investing) opportunity

This seems increasingly true. And for me, raises the question: should we compare FFH only to other insurance companies, or also other excellent capital allocators? From a Canadian perspective, CSU (Constellation Software) comes to mind. But perhaps comparisons across entirely different industry sectors are too hard.

For a number of years (well beyond the "seven lean years"), there were nothing but headwinds facing each dimension of FFH. Over the past 5+ years, it seems there have been nothing but tailwinds. Some of the tailwinds will at some point again become headwinds- this is predictable (softer markets, lower interest rates, catastrophe losses). The hope is that the prevailing winds have changed and are for the most part, at our (FFH) backs. We should also count on the appearance of unexpected (new, unpredictable) headwinds - in the past, for example, the short attack, the basis of which was entirely ill-founded (it was closer to sinking the company than many of us realized, according to The Fairfax Way). But I believe (hope) the understanding of the company by those who have carefully documented its many aspects now provides some confidence that there is (distinct from the past) a significant margin of safety. Does that summarize things fairly?

What one hopes for from a financial analyst is someone who can predict future earnings, much as we expect a true scientist to build models that help make predictions which are then tested/verified by experiments. This differs from, say, historians who try to explain the past. Some people argue that many economists have difficulty even building models that explain the past. Perhaps the analyst in question is really a failed economist who cannot even explain the past.

How can you not cheer for a team that wins a game 7 with 0 shots on goal in the second period!

With the end of the Raptors season, I now have too much free time on my hands ...

I think we have a couple of amateurs tracking it over on one of the other threads.

I may need to make an apology here and grant baseball the high ground on intelligence. Or at least acknowledge a no-decision. (I was poking more at hockey and American football, for those willing to take the bait ... you outdid me!)

Let's be honest, and simply declare basketball as the sport that every other professional league is trying to be remotely as cool as. Or athletic as. Or to have it's athletes as intelligent. I know that a Canadian is supposed to support hockey but ... why be constrained by such things?

And with more liberal use of (free) exclamation points.

Experts who have looked at the new Mythos tool and worked directly with it would be in a better position to say. From the link posted to the long blog, I'd call Mythos more than "incremental" (more than a minor advance to an existing approach), rather a next-step ... but not yet an instant game-changer. Nonetheless worrisome to the public (citizens of the world, versus investors per se) as it helps attackers more than defenders (for example, Perlroth discusses how today's major nation states have the capacity to take down each other's electrical grids by exploiting software vulnerabilities). The reason for concern is that in our existing software-based world already (before Mythos), the ability to find software vulnerabilities already exceeds the time developers have to fix them ... new vulnerabilities are found at a rate higher than existing vulnerabilities can be fixed. This imbalance has been growing over the past 10-20 years with advances in fuzz-testing tools (independent of AI). The new Mythos tool makes this worse. And, even if Mythos isn't openly released, it will be surprising if the general approach is not now replicated by other organizations/tools.

I agree, definitely beyond marketing. The blog article gives convincing details on real security flaws. So they are pushing boundaries. On the other hand: if you look at the tremendous advances in the 13 years since the introduction of the AFL fuzzing tool by Michal Zalewski while at Google -- https://en.wikipedia.org/wiki/American_Fuzzy_Lop_(software) -- and improvements such as AFL++, and then throw in orders of magnitude more computing cycles as subsidized by the LLM craze, few software security experts would be surprised by these advances. The ability to chain together multiple software exploits is not new, nor the ability to do this in a totally automated fashion (I can dig out references to such work at top security conferences 10-15 years ago already). But the continual pushing-ahead of tools that find new software vulnerabilities (so-called zero-days) is a worrisome thing. If this interests you, a good read is Nicole Perlroth's 2021 bestseller, https://en.wikipedia.org/wiki/This_Is_How_They_Tell_Me_the_World_Ends

Some years ago, when there was all that action going on about FFH being shorted in an attempt to damage the company, I recall discovering that the brokerage I was using (one of the big 5 Canadian bank's direct investing platforms IIRC) automatically opted you into having your shares potentially lent out (without compensating or notifying you) if the account was a margin account, but not if it was a cash (i.e., non-margin) account. That caused me to switch to a non-margin account, out of principle as much as fear of what might happen if my shares were lent out and then something catastrophic happened beyond the brokerage's control.