HNHacker News
TopNewBestAskShowJobs

fanzhang

944 karma · joined October 10, 2006

submissionscomments
fanzhang··on Show HN: Cost of a 51% Attack on Popular Cryptocurrencies
Rent-a-miner attacks seems to be another amusing example of when the existence of a market can break the system.

For example, voting works well in a democracy, but if we can sell votes, democracy would likely collapse because rich organizations would buy it up and control all politics. (Some say this is already true; but you know it would be way worse if votes were openly sellable).

Satoshi foresaw people trying to mount a 51% attack by buying a ton of machines, and so he went to great lengths to ensure this was unlikely using mining. I don't think Satoshi foresaw the liquid AWS-like market for instant hashing power. The ability to mount a limited-time 51% attack makes the attack literally 1000x easier than a buy-machine 51% attack.

fanzhang··on Show HN: Cost of a 51% Attack on Popular Cryptocurrencies
It's still a good proxy for how easy it is to do. The attacker could buy hardware, rent from people not on nicehash, etc.

It's like how companies' market caps are determined by the the last few trades, even though the last few trades probably represent .01% (or less) of the shares in the company.

Probably best to read the numbers a bit qualitatively. A coin that is 2% nicehashable would require substantial efforts (probably negotiating with a few private pools) to mount an attack; perhaps impossible for a not-connected miner. Wheres coins approaching the double digit percents probably is quite possible if you have even lukewarm connections. And those near 100% or above are likely super vulnerable.

fanzhang··on Top Mistakes and Learnings from My Startup: Postmortem Samsamia
Can someone with historical experience starting a company with both a friend and a nonfriend comment about this?

We get a lot of advice (e.g. pg) to start a company with friends, but we also get a lot of comments one way or the other, but without personal experience in both.

Can someone who have been on both sides talk about it? I certainly have not been on both sides.

fanzhang··on As Google Feeds Cats, Owl Lovers Cry Foul
Out of all the organizations that do bad for public policy, you're going after Google. And out of all the things you can get Google for, you're going after the latent heat of their servers?
fanzhang··on As Google Feeds Cats, Owl Lovers Cry Foul
John Deere has about 66K employees and Google has about 85K. Also, the former literally makes money by altering the environment whereas the latter is generally web-based.

I think this is the exact point the OP is talking about: Google seems big because it's bigger by profits. Also, not unrelated, some people here and elsewhere feel more emotional about Google. So take some otherwise small-town public-interest story and push it for views.

fanzhang··on EU censorship machines and link tax laws are nearing the finish line
I hope this dies just like many laws the US tries to pass on regulating the Internet in a way that promotes special interests versus that of the public good.

But as a thought experiment, if you were a lawyer tasked to argue this was publicly good, what could you even say? Content creators deserve money for creating content. Aggregators are stealing some of that value and that's inefficient but currently unenforceable? .... I just can't even think of another reason.

Maybe this is protectionism (both nationally and of existing enterprises). French publications are upset they can't get more of the pie from Google who is US based. EU publications in general are upset that aggregation theory ( https://stratechery.com/2015/aggregation-theory/ ) has taken hold and distributors like publishers are getting wiped out.

This will surely accelerate a trend that Google is already doing without monetary incentive: creating their own content and putting in on top.

fanzhang··on EU censorship machines and link tax laws are nearing the finish line
Free markets aren't always good at pricing things. For example highways wouldn't get built because of the holdout problem if there wasn't eminent domain.

There is also such thing as inalienable rights: the ability to sell yourself into slavery probably should be banned, even though libertarian agents seemingly should know the value of their own freedom.

But it also seems evident in this regulation that no one is doing that calculus. Everyone is clear on "with this regulation we gain X", but only a small fraction of people seem to be thinking "and then, people will behave in Y way and we lose Z".

With many US regulations, there are public comment periods, and those comments are summarized and published. At least when many regulations pass, people know the potential downsides. Can anyone link to the EU equivalent of this? What has the EU parliament office said about the pros and cons?

fanzhang··on GDPR for lazy people: Block all European users with Cloudflare Workers
Has anyone noticed that this is brilliant content marketing by apility.io? Generate a controversial headline, news-jack a current event topic, and then disclaim at the end to prevent the serious backlash.
fanzhang··on Op-Ed Columnist: Elon Musk, the Donald of Silicon Valley
Given that Elon Musk has been arguing with the mainstream media over Twitter recently, I'm not sure a NYT columnist has the most thoughtful view of Elon.
fanzhang··on GDPR for lazy people: Block all European users with Cloudflare Workers
For size limits, as logicians, we would think that companies would just split infinitely but that doesn't seem to be the case.

For example ACA 2012 (Obamacare) applies the most onerous terms on companies greater than 50, but not a lot of 100 person companies split into two groups of 50 to dodge it.

I think privacy is indeed along a spectrum and not binary. I certainly think that EU citizens are more concerned with Facebook and the vast trove of data they have and political irresponsibility with it than with GarethsFirstApp in the Android store handling user data well.

fanzhang··on GDPR for lazy people: Block all European users with Cloudflare Workers
I agree that the Facebook/Cambridge Analytica debacle should have been prevented. I'm not totally sure what's the best legislation to have helped that while having the minimum side effects.

As mentioned before, size limits is probably good for compliance costs; if the problem is political influence, make that a key part of the law. Making part of the law liability per privacy breach can be useful too (to deter companies from lax security that end up with them hacked).

fanzhang··on GDPR for lazy people: Block all European users with Cloudflare Workers
In legal contracts "whereas" often expresses sentiment but it's really the actual terms that matter. Having drafted a number of contracts, I feel most contracts generally have a section that approximates "whereas everyone wants things to go well and everyone to benefit..."

That's great that your company works well with GDPR. I imagine many companies will. I'm also sure that the impact on your backups could have been had without the law if you so chose.

However, an organisation that works inside the UK (EU) serving many EU paying customers (presuming here) is very different from say, Instapaper, who pulled out of the EU today because they don't make very much money from EU customers.

If we pass a regulation that says everyone who is in New York for any amount of time must pass an annual 1 hour health exam (conducted by NY state), I imagine this to be totally acceptable to New Yorkers. It correlates with good public policy: you prevent communicable diseases, and can catch health problem before it gets big. However, if this rule were to be enforced strongly, someone who might stop by once or twice a year probably is better off never coming.

fanzhang··on GDPR for lazy people: Block all European users with Cloudflare Workers
Reading the law, I only see a single exception for small companies: Article 30.1 and 30.2 doesn't apply for companies less than 250 employees.

Out of an 88 page law, 1% of an auxiliary middle of the law is carved out for small companies.

I'm not sure that counts as differential application for small companies. In the US at least, large portions of entire key burdensome laws don't apply for employers below size 50, 10, 5, etc. This does not seem to be the case here.

Does anyone know whether an official impact study on innovation was even done before its passage?

fanzhang··on GDPR for lazy people: Block all European users with Cloudflare Workers
And if getting your "act together" is a substantial cost for small companies, no matter?

The word choice almost presumes the conclusion, that data privacy rules are obvious, and cheap, and akin to just washing hands after using the toilet.

Every regulation has costs and benefits. I also would love to have better worldwide privacy at no or little cost, but the fact that people are blocking the EU shows that some companies just don't see this to be the case. And they're voting with their feet.

EU citizens should accept the fact that if they support the law, they will further data privacy protections, which are good, and they will face the music if some innovation leaves or whatever compliance costs may come with it.

fanzhang··on Mussels test positive for opioids in Seattle's Puget Sound
True. So if we presume all the opoids are from human bodily waste, we would need divide by 10% (the amount excreted out): https://en.wikipedia.org/wiki/Oxycodone#Metabolism

Probably the molecule does break up with a certain half life, either due to sun, or other animals metabolising it, etc. Then that 2.2 kg brick needs to be replaced once every two half-lives.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2991390/ Gives really hand-wavy half-life of many drugs in the environment as about 500 days, so the brick needs to be replaced every 1000 days.

Also 1 ppq is the lower bound. I think 1 ppt is quite reasonable for many methods though, so that can multiply the limit by 1000x.

All the factors above can increase the threshold amount of opoids, but here are some factors that can decrease it substantially:

There are people who dump unused drugs in, so that needs no attenuation factor.

Also, mussels are filter feeders, and oxycodone might accumulate in them substantially more (like 1000x+) than the background.

Finally, where they collected the mussels is probably correlated with sewage drains due to ease of human access, so the effective "volume of distribution" can be way less than the 110 km^3.

All that is to say, the real statistics are in the overdose calls for opoids. The headline is still an interesting cocktail-party line, but probably should be taken as a statement just by itself.

fanzhang··on Mussels test positive for opioids in Seattle's Puget Sound
Just to do the Fermi problem math for fun, how much heroin is needed to reach detectability in Puget Sound?

Assume 1 ppq detection rates (like many spectroscopic methods claim to do).

Puget sound is 110 km^3 or 1.10E14 liters of water. That's 6.05E15 moles of water. At one ppq (1E15), you need 6.05 moles of heroin. This comes out to about 2.2 kilograms.

Now in terms of doses, that's a lot of doses. But that's not a lot if a drug runner could have just tossed a single brick in to reach that concentration (or emptied a single car boot in haste).

fanzhang··on The World Wants More Camel Milk
Isn't lactose the ingredient that causes discomfort with lactose intolerant people's stomach's?

As someone lactose intolerant, I looked at the citations and none of them claim A2 milk is lactose free, as I understand it.

fanzhang··on Tesla’s Engineering Chief Takes Leave of Absence at Pivotal Moment
I hope not, but I think it's normal for newspapers to maximize revenue by playing up potentially mundane events.
fanzhang··on Google Duplex: An AI System for Accomplishing Real World Tasks Over the Phone
The store is still getting the customer, the retail rep is still getting the same call. What's so horrifying?

I see this as leveling the playing field -- upper middle class people can already afford real assistants a la Tim Ferris; this just lets everyone access it.

fanzhang··on Concorde ‘B’
Is it possible that in a regulated airline market $20k roundtrip went not into the marginal physical costs but regulatory overhead/profits/recovering fixed costs? If I showed you that a can of Coke in a Disney world vending machine was $3, and told you I could get a can at $0.30, would I have performed a miracle in the soda sciences?

$8K is a normal RT JFK-LHR ticket today in business class. This is vs $800 in economy. However, Boom proposes that the same $8K can support economy-style seating traveling at Mach 2.2 instead of mach .80. It does not seem unreasonable to me that 3x the speed will cost ~10x as much.

fanzhang··on MoviePass no longer lets you see the same movie more than once
You see it all the time in the insurance business -- when you make a claim many times insurance companies verify it. Yet insurance is still a good business.

Sure, it's a nuisance, but ipso facto a bad business?

fanzhang··on MoviePass no longer lets you see the same movie more than once
Wait, making sure they're not getting defrauded is a bad business model?
fanzhang··on Kazakhstan is changing its alphabet from Cyrillic to Latin-based
Is that right? It's not just Cyrillic, but actually Russian language? If so, that's pretty funny.
fanzhang··on WeWork to Sell $500M of Bonds in Debut Sale
A lot of other potential tenants are small anyway and have the same problems as WeWork Subsidiary LLC -- they can default and go under.
fanzhang··on WeWork to Sell $500M of Bonds in Debut Sale
It may be also of interest to note that WeWork's leases with their landlords are not signed with WeWork Inc, but instead with limited-liability subsidiaries that WeWork owns.

This means that when the cyclical tenants leave, WeWork is not left with the whole bag; instead, its landlords are.

When the tenants of a building stop paying, WeWork would default on that subsidiary, but WeWork corporate would not default.

This was hugely surprising to me when I first heard it.

fanzhang··on Google loses ‘right to be forgotten’ case
One problem is that the EU regulation is a losing fight technologically. Not that they can't affect the outcome, but that the regulations are going to be only effective in a small subset of cases.

It's like copyright. Right now, like the early days, you can search warez on Yahoo and find it. No bittorrent, etc. Once search engines got smart, with copyright, the violations just got pushed to smaller sites and decentralized protocols.

If the big search engines suddenly starts letting everyone edit search results about themselves, I bet in 5 years some IPFS / DHT / blockchain based technology that lets people permanently post and search news/rumours/gossip about other people will pop up.

Should this happen? I'm not sure, but with the nature of the Internet maybe it doesn't matter that much positively.

fanzhang··on How to Decide What to Build
The B2B market is huge, but still somewhat smaller than B2C. Look at the top tech companies by market cap today. Each is one that I've used as a consumer:

Apple, Google, Microsoft, Amazon, Facebook.

If you think the measure of dollars in top market cap companies isn't representative of companies a tier lower, this is likewise true for the highest value YC companies: Dropbox, AirBnB.

As a founder or an investor, it's important not to forget about B2B, but directionally, thinking B2C is correct.

(This is in contrast to, say, thinking of only game companies if you're a gamer; most tech company values are indeed not in games.)

fanzhang··on Productivity
What if you could be more productive and happy at the same time?

I do think it's not good to have a culture where productivity is the end goal of everything. Yet I also think it's not good to have a culture where the first time the word "productivity" is mentioned, the reaction is repulsion (I appreciate at least your honesty in admitting you didn't read the article).

Just like the drive to work-death represents the worst part of the American stereotype, the repulsion against anything that produces work or economic value to me, represents the worse of the European stereotype (or perhaps in Thiel's terms, indefinite pessimism stereotype).

Can we admit that work isn't the end goal of society, while at the same time admitting that getting more work done in less time is a good thing?

Also, on a meta-front, shouldn't we discuss which of Sam's tips work, which don't, and which one produces a lot of unhappiness as a side effect? To me that would be more interesting.

fanzhang··on Repair cafés waging war on throwaway culture
I tend to agree with the $10 phone battery example, but not necessarily the TV or fridge example. Perhaps I'm biased because I know how to repair the first, but the latter two seems like I need to take time to learn the specialized skill of diagnosing it and repairing it.

Repairability isn't a very visible or salient feature of a product. So manufacturers probably do underoptimize for repairability. They can probably do more to make the most breakable parts user accessible and have self-diagnostics.

In many of the cases where repairing makes sense, the manufacturer often does make it easy. We change the oil filter on the car and don't throw the car away. Yet if a engine bricks on a 20-year-old beater, even experienced auto repairman may recommend scrapping the car over an engine replacement -- despite the engine being a visually small part of the car.

fanzhang··on Repair cafés waging war on throwaway culture
What if our intuition is wrong about the relative costs, both economical and environmental of some repairs? In particular, just because a small part of a larger whole is broken, doesn't mean repair is the more economical or environmentally friendly act.

To take a facetious example, suppose I broke 300 random pixels of a 1000x1000 LCD display that cost $100.

On visual rough intuition, it's seems clear that less than 1% of the screen is broken, it intuitively would seem absurd to toss the LCD screen versus fix it. But automated manufacturing technology is so good that it really doesn't cost a lot to make a totally new one.

On the other hand, your time is relatively valuable, as are some replacement parts. Suppose you spend 10 hours ordering parts, soldering wires, debugging, testing, etc to get it to fully functional. Why not spend a fraction of that time at work, and another fraction of time volunteering at park cleanups or buying carbon offsets?

The numbers above may be off in one way or another, and depends on the product, your pay, and belief of environmental impact of things like throwing stuff away vs volunteering / carbon offsets.

But one point seems clear to me -- which is that if I broke what visually seems like 10% of a $100 product, the fully-burdened real economic and environmental cost is a lot more than $10 to repair it. And the lynchpin is the automated production makes new stuff relatively cheap.

← PreviousPage 4 of 7Next →