jtc331

pg_repack has some significant downsides in its implementation; I question whether it’s really a default over re-indexing concurrently. I’ve certainly not gotten that impression, and we maintain many very large Postgres clusters.

Are the containers actually full? Or just the empty containers ready to refill?

That’s nonsense. It’s extremely hard to fire people in many corps. I’ve personally seen it take years even when everyone agreed (including management) someone needed to go. And often the paperwork threshold is so high that even people producing negative work aren’t fired; you just hope they eventually move on since you’re not giving them raises etc.

Isn’t that still circular though? Knowledge of the mainstream research isn’t the same as “knowledge of what’s correct”; the person who holds the non-consensus view may have equal or higher levels of knowledge about the path that got them to non-consensus.

At some level you might reasonably expect what you described to be tautological: the person holding the mainstream view knows more about they things surrounding that view. But do they know the alternatives?

It’d be interest to compare the knowledge level the person holding the consensus view has about the non-consensus view to vice versa.

My understanding is most of this waste also isn’t as dangerous or as long-lasting.

Your math is way off. A single PV produces between 300-400 watts at peak sun. It takes ~300 watt hours to drive an EV 1 mile. A very good sunny location gets 6-7 peak sun hours equivalent a day.

So that one panel gets you 6-7 miles of EV range a day.

A typical house needs probably 15-20 panels to produces it’s needs (assuming you have storage) before EVs are even in the picture.

You literally ignored all of the parent’s points about wildly changing promises and narratives and inconsistencies to make this for into your story about this just being about an “ unhealthy mix of exceptionalism, anti-intellectualism and contrarianism”.

> cost under $20

Per-dose cost is meaningless. Now do “multi billion dollar government paid for contracts”.

You equivocated from “sharp” to “expensive” which means you’re the one that’s introducing gatekeeping not who you’re responding to.

Inexpensive knives can be sharp. America’s Test Kitchen recommends a $35 knife as their best chef’s knife.

Serial passaging is effectively all “natural” (but intentional) infection. That wouldn’t leave signs of molecular techniques since none would be involved.

Anyone claiming to disprove the lab leak hypothesis by making a claim against solely molecular engineering is either hilariously uninformed or intentionally misleading you.

Hash power in the context of potential forks in the BTC chain, not a different coin.

Financial safety is generally more about cash flow than how it’s typically portrayed, and so paying off (even lower interest) smaller loans first increases your resilience to unexpected expenses (because your cash flow is more flexible now).

There’s a psychological reason too (already commented on by others here), but there is a math reason too (cash flow).

No, it's indicative of the complexity of the system.

Please show other similarly complex systems with fewer bugs: basically doesn't happen.

ISTM the definition of title inflation should be something like “titles frequently connote a significantly higher level of responsibility than the title holder actually has”.

If the title communicates incorrect information it isn’t merely a different convention; it actively causes confusion.

> Science doesn't get much more certain than this

A well designed RCT isn’t much more certain than this?

Observational studies are hardly the peak for scientific knowledge.

And we’ll over 2x the population.

People keep saying “in secret” but that’d only be true if it happened after it were filed or some such. Giving someone an amended contract isn’t fraud.

It’s not secret. This is standard contract negotiation (sending along a new draft).

> > there’s no reason for a supporter or an opponent to leak an old draft now.

It's a big assumption that it's an old draft. It could also be the last draft circulated, and that everything since has been on concurrences or minority opinions.

> In 2018, Bloomberg Busi­ness­week published a story called “The Big Hack” that was vigor­ously denied by Apple and Amazon. Based on these denials, certain tech blog­gers became convinced that the story was false. The fact that neither Apple nor Amazon sued Bloomberg for defama­tion—despite being extremely rich, finicky, and liti­gious—made nary a dent.

No, people concluded that because there were so many holes in the Bloomberg reporting that didn't make any sense conceptually.

Anthony Fauci has been at NIH since 1968. That might be a rare example, but long careers do happen.

Options? Or RSUs?

Not if it’s a physical keylogger, which I think was the primary concern.

It's an explanation for why our anecdotal intuition concludes what it does here.

I think you're misreading the point of my comment.

Given how China repeatedly covered up COVID from early on...I think the starting point is assuming that the numbers coming out of there are not accurate.

> this level of spread would be trivial to infer from antibody tests of people traveling to China.

Who is doing that testing?

Yes, though I've only ever seen people use the term "dual writing" to refer to something at a higher-than-DB-level.

The way I've done this involves logical replication also: https://news.ycombinator.com/item?id=31087197

Frankly you won’t see many in the US either, just like you don’t see many ICE Hummers.

One does not need to have done social science research to be able to recognize obvious general philosophy of science level problems with the methods used in much social science.

I’d we take your claim seriously then we have to disallow all critiques of the replicatability crisis in the social sciences that don’t come from social scientists, but that would present an obvious new problem: conflict of interest. It’s also just an absurd requirement.

Dual writing has huge downsides: namely you're now moving consistency into the application, and it's almost guaranteed that the databases won't match in any interesting application.

Preconditions:

1. Route all traffic through pgbouncer in transaction pooling mode.

2. Logically replicate from old to new.

For failover:

1. Ensure replication is not far behind.

2. Issue a PAUSE on pgbouncer.

3. Wait for replication to be fully caught up.

4. Update pgbouncer config to point to the new database.

5. Issue a RELOAD on pgbouncer.

6. Issue a RESUME on pgbouncer.

Zero downtime; < 2s additional latency for in-flight queries at time of op is possible (and I've done it at scale).