kibibyte

I empathize with this. I went to one of those “top tier” universities and had a handful of classes where I regretted being one of the few (fewer than 10) goody good students who attended lecture, and subsequently fell asleep anyway. Over time, I realized that universities like these primarily prioritize faculty who can attract grant dollars over those who are excellent teachers.

But that said, I don’t believe this author is complaining that students generally don’t attend lecture. They’re complaining that absenteeism has increased, implying that it has increased substantially recently. And that this sudden increase in the delta is a cause for concern.

I only recently switched off the catch-all email configuration. Until then, I did what you did. But over the years, I got an increasingly annoying amount of misdirected email (possibly legitimate business, marketing, and spam) sent to all sorts of usernames, and only landing in my inbox because of the catch-all.

I now just generate masked email addresses with Fastmail. Much less noise now.

I’m visiting Vietnam for Tet, and one cultural quirk of theirs that I’ve learned is that $2 bills are considered lucky money. So much so that kids there will hold onto those $2 bills as keepsakes (though they could spend them). And thus I made my first ever trip to the bank to special order a bunch of $2s.

Something I wonder is if an uncut sheet of $2 bills would be considered extraordinarily lucky, because it’s a bunch of $2s in nearly mint condition. Or if it would be considered incredibly unlucky because those kids would have no easy way of cutting them perfectly.

I was one of the fools who installed the iOS 7 beta onto a phone that I depended on with Google Authenticator. The app had a compatibility issue with that beta release that caused it to disappear all my 2FA seeds except, very fortunately, for my Gmail. There was a bit of a ruckus about this here https://news.ycombinator.com/item?id=6112077.

Since then, I always use at least two 2FA apps at the same time.

1Password can show the whole URI with the seed, and I have used it in the past to tediously restore seeds to my other 2FA apps.

My solution to this attack is to generate random words (what 1Password calls a "memorable password") instead of something totally inscrutable. Most security question fields are long enough to accept 4 words (occasionally 5). I think it should be much harder to convince a customer support agent with "it's just 4 random words from the dictionary" vs "it's 32 random characters, do you really want me to go through it all?".

(I'm sure a determined enough attacker will eventually find an agent willing to accept the former excuse, but if it reaches that point, I think I've already lost this battle.)

I think you missed the Excel spreadsheet in the parent's link. Chase (under JPMORGAN CHASE BANK) and Wells Fargo are on that list.

Interestingly, the other 2 big banks—Citigroup and Bank of America—are missing. They both do support RTP though (which apparently is a different network operated by the same entity that backs ACH). Unsure if RTP is meant to compete with FedNow.

This is something that I still consider to be black magic to me, so this is my best attempt at describing a number of the variables.

Temperature is controlled in two ways: direct heat input (e.g. gas flame heating the outside of the rotating drum) and air flow (moving air through the roasting drum to the exhaust). It's not 1 measurement though: there is bean temperature (measured by a probe stuck into the pile of beans) and air temperature.

As far as time goes, when keeping the end temperature equal, spending more time in the roasting process means that the difference between interior and exterior of the bean are closer in temperature. When you plot air and bean temperature against time, you can derive additional information: how much energy is in your roasting drum and the rate at which the bean temperature is changing.

I'm going to preface this by saying that this is an ongoing field of research. We're still learning about what is happening in a coffee bean at various stages of the roasting process. For example, we're not quite certain exactly what is happening at "first crack" (the first time you can start to hear the beans popping), or why some coffee beans simply don't have as audible of a first crack.

We can attribute the first "rules" established for consistent coffee roasting to Scott Rao, who published some of his observations in a book in the early 2010s. Some of those rules were: (1) ensure that the rate of change of the bean temperature ("rate of rise") is constantly decreasing, and (2) prepare to adjust your roast as you begin first crack to prevent the "crash and flick" (a sharp decrease followed by a sharp increase in the rate of rise). The current thinking is that the release of moisture during first crack causes the temperature to crash, and the removal of that moisture causes the temperature to uncontrollably rise back up again. Not handling this properly often results in undesirable hollow and bready flavors; this is frequently referred to as "baked coffee".

As far as vegetal goes, that is often because of roasters cutting their roasts too short (and perhaps roasting too quickly). In this case, the bean does not get hot enough for sufficient flavor development, so it more or less retains a lot of the undesirable flavors of essentially "raw" coffee.

Note that these are "rules" instead of rules because there are a plethora of edge cases out there.

This is why roasting is really really difficult. And why even some of the best roasters out there end up leaning a lot on blends and their milk drink business.

And sorry, I gotta call out everyone who suggests this: most of your home-roasted coffee is gonna taste like ass lol. I tried home roasting a bit with a fancy setup and with a Fresh Roast. I sure saved a lot of money per pound of coffee, but I always got a fraction of the quality and the flavors were never consistent. But what I gained was insanity and the realization that home-roasting isn't for me.

You poke fun at it, but great irony is that a lot of knowledgeable coffee people started poking at all the techniques recently and we've learned that almost everything towards the showy complicated side of it are completely not worth thinking about or will have worse results. Osmotic pouring looks neat but will underextract anything that isn't dark. 4:6 is still questionable over most simpler 1-2 pour techniques. That showy thing of raising the kettle up and down a lot also will underextract.

Light roasts are great! But there are just a lot of straight-up bad renditions of them, as a result of lack of adequate training on either roasting or brewing, to the consternation of many of those coffee snobs. Unfortunately, this just happens when shops follow trends.

Roasting well in general is already quite challenging and is a lot more than just arriving at a certain bean color or temperature. Vegetal flavors are very much a roasting mistake that's being passed off as an inherent characteristic of a light roast. Combine that with techniques better suited to brewing (or pulling shots of) darker roasted, and you have a recipe for a dull, astringent, sour cup.

That being said, a sour espresso shot is always possible regardless of dark the coffee is, so I'd argue it has a lot to do with a cafe owner's willingness to train themselves and their staff to work with lighter roasted coffee.

I completely forgot that Vanguard also runs a brokerage operation in addition to their funds.

Ah, you're right! I'm now remembering that Vanguard mentions something like this for their default voting methodology, among the few choices they now allow investors to choose from.

Robinhood is a brokerage, and should be passing proxy ballots through to you for companies you hold shares in.

It's arguable whether Vanguard should allow this, and whether it'd matter at the end. Vanguard is for retail investors who want to invest their money as cheaply as possible without thinking hard about it. Delivering proxy ballots, even electronically, does have a cost, and would likely make such funds slightly but noticeably more expensive. And to what benefit? After all, most retail investors don't care to vote. If you strongly care about your shareholder voice, the better solution is to buy shares in the company directly.

That said, Vanguard now lets you select from some broader voting methodologies (e.g. Vanguard's default, ESG, or "good corporate governance"). But nothing specific like "good corporate governance, except screw Musk over/always favor Musk".

This is a bit tangential to the topic, but shareholder voting is interesting enough anyway.

* Musk will vote, for obvious reasons.

* Big institutional funds almost always vote with the board of directors recommendations. They very rarely rock the boat for a lot of mostly good reasons.

* Private shareholders with a sizable stake likely will vote.

* Retail shareholders (i.e. you and I) are notorious for not voting. Case in point: AMC is mostly held by retail investors at this point and had to go through all sorts of shenanigans to be able to increase the maximum number of shares they could issue, because that typically requires a shareholder vote that simply wouldn't ever get enough total votes to pass.

You seem to know thing or two about BART's rail infrastructure. I want to pick your brain on something.

The new BART fleet is much nicer than the old fleet in many respects, except for one very big issue: when a few drops of water fall onto the track, they have to slow down all the trains traveling through affected parts of the system. Worst case in my experience, this increases my travel time by 33%, which really adds up on the longer distance rides. My understanding is that in the rain, the new trains are just too good at applying brakes to all the wheels when they detect the wheels slipping on the tracks, and as a result, the wheels grind against the track to create flat spots. The current mitigation is to slow down the train, and it doesn't sound like they've figured out a long term solution yet.[0]

Your point about the system having a "nonstandard wheel profile and corresponding rail profile" stood out to me here. So I'm wondering, just how nonstandard is that? And is this wheel flat problem something exclusive to BART (and, perhaps a result of this nonstandard wheel profile)?

[0] https://www.nbcbayarea.com/investigations/bart-wet-weather-b...

This wouldn't surprise me, given that Amex does have a consumer reputation for being very friendly for disputes. That factors in two ways: a successful dispute means that the merchant is completely out that money, and every dispute also costs something like $25 to handle regardless of whether the merchant ultimately wins or loses.

Ah, I think we're talking about different things. I think you are talking about the overall affordability of the JR pass (and Japan tourism in general) to someone whose income is in USD, whereas I was talking about how much more expensive the JR pass is relative to the cost of train tickets (also denominated in JPY, and therefore also cheaper to people using USD), and it being possible to pay the JPY price of the pass directly without making income in JPY, despite all the authorized resellers pricing the pass in the currency of their target markets.

To give a further clarifying example, if I were planning to take the shinkansen round-trip between Tokyo and Osaka, the price is JPY 28000. Back when the JPY was stronger, the price of the 7-day pass was JPY 29650, so the pass would have been a no-brainer. Today, even though the 7-day pass costs JPY 50000, the currency is much weaker, so the USD price has increased only a little bit. However, the price of that train trip in JPY has not changed much, if at all. Therefore, because JPY 28000 (USD 184) is significantly less than JPY 50000 (USD 330), it makes no sense at all to buy the pass today if this is the only JR trip I'm planning to make.

That said, the pass is still only for those with a tourist visa (15-90 days for sightseeing), or for Japanese citizens who have lived abroad for at least 10 years.

Most credit cards can do foreign transactions, and a number of travelers have cards that can do those transactions with no fees at fairly competitive exchange rates. In the past, most of those were limited to cards with annual fees, but it's easier now to get such cards with no fees (e.g. Apple Card). Personally, I purchased the pass online from JR in JPY with a US Visa card, which converted it to a USD price that came out ahead of all the authorized resellers that sold the pass in USD.

For what it's worth, jrpass.com is a reseller. JR nowadays has an official online site to sell the pass, which prices in JPY. https://japanrailpass.net/en/

Others have pointed out the currency conversion, which helps in keeping it affordable. But it's worth pointing out that individual train fares are also priced in JPY, and therefore are also cheaper in USD because of the same currency conversion. Compared to the huge increase in price of the pass, train fares themselves have stayed relatively stable. As a result, while the pass itself may still be affordable to those of us using USD, it's even more likely nowadays that it just makes more sense to pay for the individual train fares than for the pass.

You can do it online as long as you purchase your JR pass from the official JR-operated websites (https://japanrailpass.net/en/ -> https://www.japanrailpass-reservation.net/), and not through any of the third-party authorized resellers (e.g. jrpass.com, which appears to have won the SEO game). I was able to book tickets on specific trains online quite conveniently through the same website that I purchased the pass. While I still needed to go to a kiosk to print out tickets, I feel like I reaped the benefits of it by being able to snag windows seats early on a few scenic yet nearly full trains.

> Let's hope the HSM supports secret backup well enough

Secrets backup for HSMs from a certain vendor—that you may or may not have named in your comment—I've worked with is actually the easy part. You just make copies of it and all the key data and check it into a git repo, because all of that data is protected by an HSM secret. Distributing that HSM secret among several HSMs for redundancy is also pretty easy.

The hard part is all the administration around it, specifically around custody of the smart cards that contain chunks of the HSM secret: where are they protected, where are the backups of the cards, who has access to them, coordinating sufficient card custodians to meet quorum, etc. You need to meet quorum to provision HSMs with the same secret.

The real "trust me" part of this is arguably less that the vendor backdoored the HSMs, and more that Apple pays the vendor support contracts (that hardware eventually fails) and maintains the knowledge continuity for the teams responsible for administering these HSMs as people join or leave those teams over time.

For what it's worth, this is pretty much why you don't see HSMs used often at less mature companies.

I just started the Japanese course recently and was trying to figure out why a bunch of the criticism I read of Duolingo's course seemed irrelevant or inapplicable. I guess this pivot explains why to some degree.

Yep, this is "level 3 data". Merchants or payments services often get lower interchange rates for providing this extra data

I'll torture this DNA and C source code analogy a bit.

Epigenetics is missing in this discussion about reproducing a human from just the DNA. These are superficial modifications (e.g. methylation, histone modification, repressor factors) to a strand of DNA that can drastically alter how specific regions get expressed into proteins. These mechanisms essentially work by either hiding or unhiding DNA from RNA polymerases and other parts of the transcription complex. These mechanisms can change throughout your lifetime because of environmental factors and can be inherited.

So it's like reading C source code, except there so many of these inscrutable C preprocessor directives strewn all throughout. You won't get a successful compilation by turning on or off all the directives. Instead, you need to get this similarly inscrutable configuration blob that tells you how to set each directive.

I guess in a way, it's like the weights for an ML model. It just works, you can't explain why it works, and changing this weight here produces a program that crashes prematurely, and changing a weight there produces a program with allergic reactions to everything.

There are surge protectors that can protect the coax cables used for cable internet. Here's one that I use: https://www.amazon.com/Tripp-Lite-Right-Angle-Protection-TLP....

I'm curious when you bought that TI calculator and whether you were aware of the whole scene around the TI-83+/TI-84 calculators. When I was in high school, these calculators were the required calculators for certain classes and exams. But a lot of my classmates liked these calculators because of a niche online community that had figured out how to write assembly programs that targeted the Z80 processor in these calculators. And that meant that we could load games like Tetris or Phoenix (a Space Invaders clone).

It's cheap because it doesn't come with a case or a PSU, and appears to only be available through AliExpress (not necessarily a bad thing, but could make manufacturer support tricky).

A case might be nice if you prefer not to have a bare GPU on your desktop and if you want to slow down the dust buildup. SFX power supply units appear to run you about $100 for something like 650W. The power delivery is limited to 60W (in terms of how much power it'll deliver to charge a laptop), which is sufficient for many laptops but probably not all of them (I think the last generation of Intel MacBook Pros might want more than 60W).

This was part of my childhood (almost 20 years ago!). Growing up, we had a lesson in either elementary or middle school in which we were all asked to read this website on the Pacific Northwest Tree Octopus and fill out a research assignment on it. At the end of it, the teacher asked us if we thought it was real and revealed that it wasn't, with the lesson that, yes, even though the internet is a valuable research tool, that we shouldn't believe everything we read.

As a bonus, I think we were also asked to review some information on the aluminum foil hat. https://zapatopi.net/afdb/build.html

I'm glad to know that the ferry exists, and it does look like that it's common for them to offer special event service. My main gripe though is that the special event fare is rather steep for what it is ($36.50 round trip). Sure beats driving up there, but I'd almost certainly be weighing other interesting things to do for that price.

Oh interesting. I remember that Folding@Home way back then (ca 2009) was already testing protein folding on GPUs and it took advantage of CUDA. I never really thought much of it other than how cool it was that my mid-tier Nvidia graphics card could be used for something else other than games, but this explains how this ended up happening.

(Bit of a tangent, but that project was very influential in getting me interested in computer science because, wow, how cool is it that we can use GPUs to do insane parallel computing. So I guess, very very indirectly, Nvidia had a part in me being a software engineer today.)