conjectures

Sure, but why start off by trying to seem nice? Can only assume they've been wasting time interviewing people likely to be fooled by that level of bs. Or is it that they don't have a clear view of what level bs is required to fool people? Either way it shows a reliance on bs and some flavour of incompetence.

'Something, something, hiring feels like a referendum on whether candidate is good enough to join a fancy club... ...FAANG have a straightforward entrance exam... ...on the other hand Honeycomb's club is smaller and more exclusive.'

Wait, what's the point here?

IP aside, anyone else concerned about the business of HiQ?

I presume what they are doing is:

* Scrape profiles.

* Calculate time delta in jobs.

* 'Predict' churn rate for (prospective) employee.

With respect to prospective employees in particular this seems likely to entail lots of risks. Average job time delta is going to be a massively overdetermined variable, and noisy wrt 'next job delta'. I'm worried how they're going to sell that to employers.

Let me translate that for you:

“Independence is a strong part of the character of the people of the island. We’re whatever it takes to remain a tax haven.”

With energy bolts hammering at her heels on every step, our plucky lawyer dived through the doors of the high court. Slamming the injunction docs into the outstretched arms of a waiting clerk she yelled at the shimmering blue spheres on the other side of the window, "You've been served."

Love the idea that if an opposing power developed force fields + compact fusion reactors, that the last line of defence would be patent law.

> Their best hope is a cultural revolution.

You say this phrase, but I do not think it means what you hope it means.

'No credibility' suggests that he was not telling the truth. However, the facts you point out would also be consistent with someone who was motivated by revenge to tell the truth.

I know what you're getting at, and there is something there. However this argument is unfalsifiable.

No society ever collapsed due to policy on repairing washing machines. One could survive indefinitely in a sub-optimal policy regime on a topic of limited importance without collapsing.

This is a great excerpt.

Sounds like this guy got pipped to publication by Peter Godfrey Smith's Other Minds.

Can anyone who has read both comment on whether Graziano's book contains much different?

That sounds like something nation states would be able to do well before it's acknowledged.

Another way to look at is that children have the right to not have their health jeopardised by the fact free opinions of their parents.

Surely because the thread is about people wondering, 'why are monasteries declining?'

A substantial reduction in religious belief could explain away a large part of monastic decline.

I don't see anything in these statements about the usefulness of academic macroeconomics, except insofar as certain actors believe it. Is that right?

Well, he's not wrong.

It's a system which is slow to change, nigh on impossible to experiment with, we only have one experimental unit, and it responds to interventions. No wonder it's difficult to science it properly.

'Macroeconomics' by Williamson is my go-to example of the worst textbook I ever had to read (setting aside postmodernism). Overly certain, badly argued, with fishy math. So, a good place to go to for everything wrong with the discipline.

'Wire-service' as C20 jargon for copy pasta.

"number of datapoints, not the size of the inputs"

For brain imaging, 100k was 'the dimensionality of the input' (but also the number of datapoints). This could easily be O(10^6) though with higher res imaging.

I'd check out, for discussion:

https://www.prowler.io/blog/sparse-gps-approximate-the-poste...

For code:

https://github.com/GPflow/GPflow

Based on the people involved knowing what they are talking about, rather than on experience using that particular work.

First one I tried with GPT2

In: This is a thing I hope to Jesus

Out: I am a whore\n\nI am an

The real issue is that inversion / solving is O(n^3).

This really hammers you for high dimensional settings (e.g. brain imaging n>100k).

Solutions:

* Pick a very special kernel with an known / approximable solution.

* Reduce the dimension of the problem (sparse gps).

Skilled trades are still a good idea in the UK. I estimated that experienced self employed electricians could probably pull in $80-120k per year excluding office costs at 80% utilisation.

It's not huge money, but once you start factoring in that tax is much better for the self employed, costs of living, commute time etc it does look like a good deal. Even if you discount the estimate by 1/3 it still compares very favourably to most white collar careers.

I get that, and you may not be wrong. It's that you may be right about the wrong thing. All this stuff matters a lot if you're pursuing a career in academia. But from the outside it just means the machine of science is a bit less efficient than it could be.

Also I put it to you that serious researchers are rarely misled by low quality, vacuous papers. That kind of work will be ignored by history.

So I agree that what you're pointing out is something of a problem, but may not constitute a cataclysm.

It's called herd immunity, and they're undermining it.

> they shacked themselves with decades of debt and get a mortgage

A mortgage is just renting a place from the bank at the market price of when you bought the house + some enforced savings. It's hardly a mill stone compared to renting for the same number of years.

On the other hand, to list some random newish things:

* Genome sequencing.

* New malaria drugs.

* Graphene.

* Better batteries.

* Deep learning.

* fMRI.

* CRISPR.

From the outside, the public doesn't care about the trials and tribulations of PhDs. Or irrelevant puffed up papers. Or the organisational politics of research.

Massive misundertanding here:

Science doesn't work by noble individuals scrupulously following rules that they could ignore if they chose to.

Where science works, it works by incentivising one scientist to tear apart the next scientist's position. There's no better way to avoid having your argument torn apart than for it to be genuinely solid.

This is aptly demonstrated by Nature tearing into Elsevier...

The time to get really worried is when you stop hearing anything like this and get a message only of, 'Everything is fine, nothing to see here'.

I'm afraid you're mistaken. Randomisation allows one to make the strong causal assumption that the treatment regime allocation is unrelated to any of the other variables, observed or unobserved.

Anyway, the section you're pointing to agrees with me. It just happens to be overlooked when they summarise...

> It is well known among practitioners that not all models have equal learning capacity and that additional data is wasted on those with lower capacity

Not just practitioners, IIRC there's a whole bunch of theory under the handle of Vapnik-Chervonenkis dimension.

Welcome to numerical computing :D!

Btw:

julia> (-80538738812075974)^3 + 80435758145817515^3 + 12602123297335631^3

42

I mostly agree with Pearl.

However, the account of traditional statistics given in this article is misleading. Randomisation is a major part of traditional stats, and it is inherently a causal hypothesis: breaking the links between unobserved covariates and treatment regimes.

An important alternate contemporary causal inference framework by Rubin has origins in a 1923 thesis...

...but the content of Pearl's approach seems superior; if you ignore the academic spats.