Hacker News Reader

suchireon Driving engineers to an arbitrary date is a value destroying mistake

I think another rule of thumb I would add, which other commenters have mentioned in passing, is to negotiate scope, quality, and likelihood of success, not estimates. Negotiating estimates is where people start questioning each others’ expert judgment, which leads to a breakdown in trust.

suchireon School districts take unplanned plunge into online learning

We do, but generally access is scarce and so many are homeless

suchireon School districts take unplanned plunge into online learning

These are likely insufficient for video-based schoolwork or materials. That would max out any quota set for these hotspots

suchireon N95 Respirators vs. Medical Masks: A Randomized Clinical Trial (2019)

Cochrane reviews (of which this is one) are the gold standard for meta-analysis and evidence-based medicine. I trust their work a lot

suchireon U.S. will suspend all travel from Europe for 30 days

Not sure the CDC is the most credible source at this point

suchireon As the Startup Boom Deflates, Tech Is Humbled

It is rare for innovation to be technical or purely technical. It is far more common for innovation to be from a novel business model, which perhaps is enabled by the technical innovation from someone else. Many times one technical innovation (Internet, smartphones, PCs) leads to many business opportunities (Amazon, Uber, Microsoft)

suchireon Where Sci-Hub Is

Scientists don’t get paid anything regardless of whether or not you pay a subscription/article fee to the science journals. The best thing to do to benefit scientists (usually) is to spread awareness of their work, as that is the main currency in academia that determines career advancement and future fundraising. Ideally you’re promoting it to other academics (so that they generate trackable citations), but general publicity also helps.

suchireon I've screwed up plenty of things too

You do it by management not making things zero sum, and by correcting/penalizing people for blaming behavior (e.g. saying a root cause is that a team/person made a mistake, vs a system not preventing or mitigating the mistake)

suchireon Ask HN: How was your first performance review?

”But in that case, usually they are already giving you feedback during the process and a formal performance review is just a checkbox item to satisfy their own supervisor or funding organization”

If this is ever not true of your manager at any company in any industry, then the manager sucks and you should leave (if you have the means). If the manager wants to do this but the company doesn’t let them, then the CEO suck and you should leave.

suchireon Cormac McCarthy on how to write a science paper

It’s definitely possible, but I think there’s still a palpable difference today. As an example, this work from 1989 on how telomeres work was awarded a Nobel in 2016. It’s not unreadable, but it has a completely different (and I would say worse) style compared to the Crick paper I linked earlier: https://moscow.sci-hub.tw/1036/810d9cdd2ea2e1a07d887dc432061...

It’s also interesting how formats have evolved. Older journals like Nature and Science used to have a “letters” section--essentially letters to the editor. They were brief (200-300 words) and started with “Dear sirs”, like an actual letter. Today, a “letter” in the same journals is essentially a whole (short, 2-3 page) paper, like this one: https://www.nature.com/articles/s41586-019-1578-4

suchireon Cormac McCarthy on how to write a science paper

One of my PhD classes was a paper reading seminar that spanned most of the history of modern molecular biology, from 1959 to the present. It was starkly apparent how much worse the writing quality became around the mid-1980s and beyond. Compare the classic [“genetic code” paper by Francis Crick](https://moscow.sci-hub.tw/1024/2cfa4b970e4a1fd816ee18afa95aa...) with most modern papers

More science authors should read writing advice like this.

suchireon Google Photos is making photos semi-public

I used to be the tech lead for the sharing and permissions side of a file storage service. In my experience with designing systems, participating in user studies, trying to problem solve with coworkers, and so on, this is an extremely hard problem to solve, because (as can be seen in the comments), there usually isn’t a right answer. Different people expect very different defaults, and get frustrated and upset when things don’t match their a priori expectations. The temptation is to solve these differing use cases with lots of configurability, precise descriptions, and lots of user education but users don’t read (usually), and rarely do they understand the tricky implications of different settings. Not only that, but mistakes they make in configuration or use will understandably cause them to be more scared of using your service.

I don’t think anyone has solved access control, not Facebook, Google, OneDrive, or even Apple.

suchireon Gmail’s API lockdown will kill some third-party app access, starting July 15

1) Maybe a solution could be to give power users an escape hatch with big red warnings that doing so voids any support or terms of service guarantees, sort of like how Apple allows non-signed apps, but not be default.

2) Even with a mechanism like the above, part of the issue with communication/social data like Gmail is that you can’t control your own privacy perimeter, because the other party could inadvertently leak your data through another app they connect

suchireon No Kings: How Do You Make Good Decisions Efficiently in a Flat Organization?

The way my workplace addresses this is:

1. Transitioning to management or tech lead is not a promotion, but a change in responsibilities (and transitions back are common);

2. Whether manager or IC, we only promote people based on demonstrated ability, not potential, to avoid the Peter Principle

suchireon No Kings: How Do You Make Good Decisions Efficiently in a Flat Organization?

Large companies tend towards politics because it’s hard to build empathy and trust across thousands of people. Academic departments tend towards politics because nominal leaders either have no training or no power

suchireon No Kings: How Do You Make Good Decisions Efficiently in a Flat Organization?

At least at many big companies, there’s a dual career ladder for ICs and managers, with equal pay across equivalent levels (e.g. Manager ~ Senior engineer, Director ~ Principal Engineer)

suchireon A guide to difficult conversations

I also highly recommend the book “Difficult Conversations”, which has a similar overall idea with a slightly different way of framing and breaking down the hard parts of these conversations.

suchireon Ask HN: Favorite books on how to be a good manager?

My favorites so far: High Output Management, Turn the Ship Around, Radical Candor, and the Five Dysfunctions of a Team.

suchireon Software architecture is failing

The kinds of things I look for in evaluating senior candidates are things like:

- Payments flow kept going down because of lack of reliability in our CI, so I rebuilt the job queueing system to get to 99% reliability

- Most of our business is outside of the US, so we invested in better CDN routing, got a 50% improvement in our web TTIs

- We weren’t able to deploy code often because of our monolith, so we moved to a service-oriented architecture in order to speed up development

If candidates can’t justify why a project happened and why it helped either the company or their team, then that candidate isn’t a senior engineer, they’re just someone taking marching orders.

suchireon Exactly-once or not, atomic broadcast is still impossible in Kafka

The older choice usually by default has:

- More people who know how to operate it

- A greater portion of its failure modes that are known, with workarounds

- Mature tooling and libraries

New technologies need a compelling reason that their pros will outweigh the above incumbent benefits.

suchireon Show HN: Add a play button next to IPA on Wikipedia

Likely "some" as in "home", not as the real English word "some"

suchireon Code is not literature (2014)

Reading code may not be like reading literature, but I find writing code to be remarkably similar to writing prose. I wrote about that in more detail here: http://www.ericsuh.com/blog/posts/2016/01/writing-code.html

suchireon H-1B reduced computer programmer employment by up to 11%, study finds

From the abstract of the paper itself: "In the absence of immigration, wages for US computer scientists would have been 2.6% to 5.1% higher and employment in computer science for US workers would have been 6.1% to 10.8% higher in 2001".

So focusing specifically on US workers, not immigrants.

suchireon Why Is It So Difficult to Develop Drugs for Cancer? (2010)

Those are classic success stories mainly because they were the low-hanging fruit for targeted cancer treatment. For blood-cell cancers like leukemia and lymphoma, treatment has come a long way due to advances like the ones you mentioned.

For "solid" tumor cancers like breast, pancreatic, prostate, and skin cancer, we're still struggling to make a dent in the death rate.

suchireon Dropbox Is Struggling and Competitors Are Catching Up

Dropbox does support a paid business account and a personal account. The paid business account just needs to be a Dropbox for Business account.

suchireon As a math professor, how can I help grad students find nonacademic jobs?

The best thing is to give your students the freedom and encouragement to do internships during their PhD/MA/MS. That'll help them find outside mentors, experience, and contacts, as well as in-roads to future employment

suchireon Ask HN: What are the best product landing pages you've stumbled upon?

That was from a previous redesign. There is a newer landing page within in the last month.

suchireon How Google Hires

+1. People who have been in management (or are "people" people) are often good speakers that do very well at a surface level, so it requires digging deeper to get a richer picture of what they were thinking and what they did.

suchireon The Terrible Technical Interview

IMO, having interviewed hundreds of candidates, whiteboard coding selects works very well for hiring a very specific type of fresh-out-of-CS-undergrad student, but it breaks down in almost every other scenario.

For more experienced candidates, we've swapped out most of the whiteboard questions for several other types of interviews, including in-depth technical discussion of one of their past projects, a discussion of how to architect a non-trivial system, and a more soft-skills interview on how they work with others, break down projects, and so on. This gives us a much rounder picture of a candidate, and allows candidates who are very good at something to shine a bit more.

suchireon Adventures in message queues

Note that even in CP mode, committed messages can be dropped as well (one of the pitfalls I assume you've experienced) when the cluster heals from partition. cf link in my sibling comment.