Hacking Startups

What is Bitcoin good for? Not Silk Road.

Tue, 25 Jun 2013 18:28:00 -0700

Bitcoin is uniquely suited for applications where capital mobility is critical. Other applications are largely spurious.

There are several proposed advantages of cryptocurrencies in general and Bitcoin in particular:

Anonymity
Store of value
Medium of exchange
Investment
Irreversibility
Independence from any government policy
Freedom of capital movement

We examine and compare, for each of these purposes, as potential alternatives, several existing currencies and stores of value:

Dollar cash
Gold
PayPal
Dollars in a domestic bank
Dollars in an offshore bank
Land
Shares

Anonymity

Widespread confusion about the potential anonymity of Bitcoin is understandable. Theoretically, cryptocurrencies can provide stronger anonymity assurances than alternatives (see, e.g., Zerocoin). However, Bitcoin provides no anonymity, not by design and not in implementation. Transactions are globally visible and traceable; they can be tied to specific people when money is spent on tangible purchases. True, you could stuff some Bitcoins into a pseudonymous wallet for a while without being traced. You can also open a PayPal account without verifying it and get some money into it.

The benchmark is not whether you can anonymously receive money, but whether you can anonymously spend it.

If all you worry about is being able to anonymously receive, I have a better idea: ask to be paid in empty eggshells and have them deposited in the closest body of water. You’ll never get to them, but that only improves security.

Store of value

Bitcoin is also an exceedingly poor store of value compared to gold or land.

Gold has remained a relatively stable store of value for the last five thousand years. Land has even appreciated, largely due to population increase.

Given how long these trends have continued, it is likely that they will continue for at least some fraction of the time that they have already continued. I do not recommend gold as a store of value, but it sure beats Bitcoin, which has a much shorter track record, much higher historic volatility, and much more uncertain future. It is only reasonable to use Bitcoin as a store of value if your time horizon is a few months; that is to say, never.

Medium of exchange

Bitcoin has no real advantages as a medium of exchange. It is not faster than credit cards or wire transfers remotely and the difference in fees is largely due to the fact that banks provide additional services—most importantly, various forms of protection from fraud and theft. In person, cash works better.

Investment

The investment function of Bitcoin is similar to store of value, but distinct. While Bitcoin is an exceedingly poor store of value, it may be, for the right market participant, a good medium for speculation. Whether you are such a person depends on your sophistication as an investor.

A simple benchmark question is: Do you have more than a billion dollars under management?

No? Well, pal, then you are meat for those who can move the market. Given that Bitcoin is a small market in total, the trading trends will be for the conceivable future dominated by manipulation.

See also below on the effect of zero supply elasticity.

Another category of trader for whom Bitcoins can be great for speculation is an insider. (Un)fortunately, most insider knowledge about Bitcoin (a bug or perhaps future criminal prosecution of large exchanges or software makers) will need an effective way to take a short position. Due to counterparty risk, Bitcoin today does not offer a good way to short itself. This is an area that a different cryptocurrency could address, perhaps through an inherent mechanism of futures contracts settlement.

Irreversibility

Many tout irreversibility of Bitcoin as an advantage. Both cash and gold, however, have a very comparable extent of irreversibility.

Perhaps the most important point missed here is that reversibility is a feature, not a bug. Markets vastly prefer settlements reversible in case of theft or fraud; these protections increase consumer uptake.

Liability limits on credit cards enabled their use on the Internet despite initial consumer fears. Credit card issuers have improved reversibility by lowering the legally mandated $50 liability limit to $0 on most cards, further increasing uptake.

Consumers don’t want to lose the contents of their accounts with no recourse because of a security breach on their computer. Reversibility protects them, and they want it.

Independence from any government policy

Here is where we are starting to get somewhere. Government could pass all sorts of laws regulating or even prohibiting Bitcoins, but it lacks technical means of just grabbing them like it can with contents of domestic bank accounts (think Cyprus) or devaluing like with cash (think everywhere).

However, gold has all the same properties here that Bitcoins have, and is a vastly better store of value. (Again, I do not recommend you take any position on gold, and in particular I would not go long at the time of this writing. I am simply pointing out that it is an existing alternative structured similarly to Bitcoins and just plain better for the stated purpose.)

Note that both gold and Bitcoins are better at this than cash, which can be simply declared invalid, domestic bank deposits, which are in even far greater danger, or something useful, but also highly visible, like real estate, farmland, or other large tangible assets. Gold and Bitcoins can be made illegal, and population can be ordered to hand them over, but the order does not implement itself, like with voiding cash, and is not trivial to implement, like seizure of domestic bank funds. And, obviously, offshore banks are domestic banks in a different jurisdiction. The practical ease of affecting physical gold and Bitcoins with policy is comparable. Bitcoins can be hidden using physical and also steganographic means, whereas gold does not benefit from steganography; in practice, however, this difference is small, unless one wants to move the store of value. Which brings us to freedom of capital movement.

Freedom of capital movement

Finally! Bitcoins are like an offshore bank in the Internet crypto jurisdiction.

Once some local currency or other form of value is in Bitcoins, no technical means can stop the capital from leaving an oppressive regime for any other place.

While Bitcoins may be subject to all sorts of dangers in the longer term, the move does not take long, and if other alternatives are not available (e.g., it is illegal to carry gold or cash out or make wire transfers out of a country), Bitcoin transfer can only be stopped at the point where the value is converted into Bitcoins.

Zero supply elasticity means higher volatility

Bitcoin has a unique and undesirable property of having no supply elasticity. More gold can be mined and more dollars can be printed, but Bitcoins get minted on schedule that ignores demand. Zero supply elasticity means an inherently higher price volatility compared to gold or dollars.

Rarely considered attacks

A security assessment is always relative to a threat model, so I will enumerate here a few insufficiently considered attacks on the Bitcoin ecosystem:

Make Bitcoins and any dealings with them illegal. Treat them legally as the money laundering equivalent of child pornography. Pursue any exchanges, cutting off their transfers and attempt to seize deposited funds; make creating, offering, and running Bitcoin software illegal.
Simply have more than 50% of total compute capacity and rewrite history to confiscate other people’s Bitcoins. Current system capacity is about 150 Thash/s. Current best price of mining hardware is about 50 Mhash/s/$. That’s just $3M worth of hardware. Given the total valuation of Bitcoin, this is highly unstable (due to the temporary nature of availability constraints of ASIC miners) and will change in future, but even then, you can always spend the total valuation of Bitcoins for the ability to rewrite history. Even at graphics card efficiency (1.5 Mhash/s/$), this is still only $100M, potentially profitable even for simple theft. Even at current point of equilibrium (~$1B), this is laughably easy for a large government agency with crypto expertise.
Quietly pressure Bitcoin software makers to do nefarious, but seemingly innocuous things, like purposefully insert bugs that allow for a fork. There already was a fork.
Adopt Bitcoins for normal finance purposes, but with a few protocol changes that will evolve gradually. Final result will be no better than current banking system in terms of government control, but each step will be relatively benign. An easy way to start is to provide limited reversibility to be able to offer fraud and theft protection.
Fork the ecosystem rather than the record of transaction history. Release a different, and slightly better currency. E.g., a version of Zerocoin that is not Bitcoin-compatible. Rinse, repeat.

Conclusion

Bitcoins are great if you need to do a small move of capital (something of personal scale for an average person) that you can’t do otherwise. Once the move is complete, exit the Bitcoin position.

If you are a cryptocurrency researcher, a fruitful and challenging area, besides the obvious anonymity, are built-in derivatives, particularly futures and leverage.

Local development branches in Git are a bad idea

Wed, 04 Feb 2009 17:00:05 -0800

Git makes it possible for a programmer to do most work on a branch that’s not pushed to any repository coworkers see. This is a bad idea. The only justification I’ve seen was that the practice “let[s] you do some bonehead moves without your co-workers ever having to know”.

You should not need to hide your code from your coworkers. Even your experimental code.

Local development branches conceal information about the history of the codebase. The purpose of version control is to preserve this information. Using a local branch defeats part of the purpose of version control.

If you’re going to make boneheaded moves that break stuff temporarily, making a branch makes total sense. Just push it out where it’ll be preserved and visible.

Choosing a web dev platform: Rails or PHP?

Fri, 16 Jan 2009 02:18:54 -0800

When building your next web app, you can create value by improving existing web dev tools or by making a compelling app. If you’re interested in making a great app, it’s best to use the best of existing tools and let others improve them.

Most web pages served today are produced by one of these platforms:

Static HTML (brochure-style sites)
C or C++ (search engines mostly, some backends)
PHP (most high-volume apps)
Perl (legacy high-volume apps)
Java (many high-volume apps)

In addition, a large number of new apps are written in Ruby on Rails. It does not account for a noticeable fraction of page views, but it does account for a large fraction of innovation. Then there’s Python.

Since you’re building an app, you can’t use static HTML.

Perl as a webdev language is obsolete. It used to be the default during the infancy of the web, and large apps continue to run smoothly on Perl, but it should not be used to build new ones.

C should not be used for an initial implementation. Instead, it can be used when the app needs to handle huge volumes to rewrite the critical bits.

Java is a matter of cultural fit. If you say the words “enterprise software” with a straight face, plan to run Oracle on Solaris, and outsource good chunks of work, you don’t need me to tell you to use Java—you probably aren’t even considering other options. For most web apps, however, and for almost any startup with technical founders, Java is a poor choice.

This leaves the choice at— Rails or PHP?

If you need to integrate with something in PHP, go with PHP. For example, Facebook apps should be written in PHP by default.

Rails allows programmers to work much faster, and is more pleasant to deal with. If you need to hire, better people will be available to work on Rails. This is because Rails is very high level. The downside of high-level tools is performance. It’s not just slower. It’s much harder to make faster.

PHP runs about 40 of the top 100 sites. It’s the default no-brainer choice, lower-level and uglier than Rails, but perfectly workable. It’s easy to hire for, and easier to make fast.

To choose, consider these factors:

Size of app, particularly UI: large apps are easier with Rails
Size of team: small teams can do more with Rails
Quality of team: PHP is easier to work with
Need for change: changes are easier in Rails
Expected initial traffic: high-volume apps are easier in PHP

Unless learning Rails looks hard or there’s a huge marketing budget in place at launch, consider Rails.

"Gap, Victoria’s Secret and J. C. Penney are experiencing double-digit sales growth at their shopping..."

Mon, 21 Jul 2008 20:03:40 -0700

“Gap, Victoria’s Secret and J. C. Penney are experiencing double-digit sales growth at their shopping Web sites”

- To Save Gas, Shoppers Stay Home and Click - NYTimes.com

"[Osama bin Laden] said at one point that he wants oil to be $144 a barrel"

Thu, 03 Jul 2008 22:20:49 -0700

“[Osama bin Laden] said at one point that he wants oil to be $144 a barrel”

- Fears, Again, of Oil Supplies at Risk - New York Times, October 14, 2001

10 useful Facebook apps

Thu, 03 Jul 2008 20:18:00 -0700

Most Facebook apps are giggly. But serious apps exist even today when they cannot spread.

A comment on my earlier post 75% of Facebook users are giggly and poke; 25% are serious and import bookmarks asks if there are, indeed, any serious Facebook apps. Here are some of the serious apps that I have installed:

Analytic Polls – opinion comparison/sharing.
Antlook – news. Personalized recommendations from AI.
Causes – charity. The only popular serious Facebook app.
Dopplr – business travel. Tells when contacts are in the same place as you.
Feedheads – tool for importing Google Reader shared items and RSS in general.
Mento – bookmarking tool that tells you when friends click. Firefox extension that takes screenshots.
Neighborhoods – meet your neighbors on Facebook. I’ve become friends with two great people this way.
PayPal – fundraising.
SeenThis – news. All WSJ content for free.
Truemors – tips and rumours.

You’ll notice that no-one uses these excellent apps, with the exception of Causes. Because serious apps are not used, developers have no incentive to create more. If my proposal were implemented, there would be more, and better, serious apps.

But before then, you can still enjoy these apps that developers made against their best interest and judgment.

"[O]ne-third of adults said high gas prices had made them more likely to shop online to avoid..."

Sun, 29 Jun 2008 17:17:34 -0700

“[O]ne-third of adults said high gas prices had made them more likely to shop online to avoid driving.”

- Envisioning a world of $200-a-barrel oil - Los Angeles Times

Would you work with micromanaging boss, no salary, and all your work thrown away?

Wed, 18 Jun 2008 21:49:00 -0700

If you think the coming nuclear winter will make job market tough for employees, you need to hear about the job offer my daughter got recently.

The job has:

$0 salary and no equity (you’re supposed to be compensated in experience)
no benefits other than vacation and sick time: no insurance, for example
no possibility of promotion or raise, ever
no job description: just do what you’re told
micromanaging boss asks about project status every hour
strict hours, starting at 8:30AM sharp
if you’re late even a few minutes, your boss sends you to her boss
rigid workweek, but then you’re expected to work from home a ton
open-desk seating, not even a cube, with a hard chair
the work is boring and demeaning, like adding digits and copying text
all your useless work gets thrown away
if you want to use a computer, you can buy one or just scribble on paper
no supplies room, either
my daughter can’t drive, so commute was complicated
can’t even put the job on your resume until you work there for a decade

I wish this was a joke or I was making it up.

Having consulted with me, my daughter of course rejected this ridiculous offer and is now just working on side projects while looking for a better opportunity.

But millions of other seven-year-olds accepted identical offers.

75% of Facebook users are giggly and poke; 25% are serious and import bookmarks

Mon, 16 Jun 2008 23:02:00 -0700

Facebook users fall into two categories: giggly and serious. Facebook user experience could be better if Facebook took this dichotomy into account. Ignoring the difference pushes apps to be giggly and serious users to be unhappy.

Giggly 75% like pokes, quizzes, pic forwarding, fun games, selling friends, glitter on profiles. They express themselves through style and interact with friends using the mouse.

Serious 25% like bookmark import, utility apps, discussions. They express themselves with text and pictures containing them and interact with friends using the keyboard.

Because you’re reading this, and made this far, you’re serious. (Giggly users tend to not read much at all, certainly not blobs of text, and quite certainly not my blog.) Let me tell you a few things about the giggly majority and propose how to make Facebook better for both giggly and serious users.

Giggly users are

younger
less educated
lower-income
less likely to have credit cards
far less likely employed
more suburban and rural
more frequently female.

Giggly users love to have fun with their friends, love to chit-chat and giggle, forward things easily and without a second thought. Giggly users generally don’t review applications because it requires typing. They don’t visit the about pages much. The prototypical giggly user is a female teenager who might later go to a party school to major in English.

You’re far more familiar with the serious users. Serious users are Harvard students, Silicon Valley types who use Facebook for professional networking, young professionals, etc. Serious users vote in primaries, care about privacy, understand the importance of financial planning, and are somewhat hesitant about the exposure of personal information by social networks. Serious users dislike the apps, infrequently use them, but write many of the reviews. Their notion of fun on Facebook is Scrabulous. Serious users are a bit boring. They make up for it by extreme sports and odd personal styles. The prototypical serious users are you and I. (Facebook employees and shareholders are also serious Facebook users.)

Problems:

Giggly users send a bunch of giggly communication to the serious users, for whom it’s annoying noise that drowns the signal from serious friends
Giggly users want more self-expression tools, which Facebook won’t create because of concern about serious users, who will hate them and cry “MySpace”

Solution: Have each user and each app self-elect into giggly or serious categories and treat them differently.

API calls returning list of friends, friend selectors, etc., should, by default, only return giggly friends for giggly apps and serious friends for serious apps. There should be a way to override this with some difficulty and user involvement.

Giggly users should be given tools to create different backgrounds for their profiles with different text colors and an option to play music on load. The default text color for giggly users should be pink, on a purple background with starbursts. Facebook should partner with RockYou to enable displaying the profile owner’s name in her favorite style of glitter, and a larger font. Latest photo album should start playing as a slide show on load. Applications should be given hooks into these extra self-expression tools, allowing iLike to set the song to play on load, etc. Self-expression should reign supreme.

Serious users should continue to see profiles of any users, including giggly, as they do today, minus the app boxes. Utility and uniformity should be emphasized.

The division of the app ecosystem will be particularly valuable. Giggly users will continue to have their silly apps, but the silly apps will stop bothering the serious users.

This will create an opening for serious and useful apps, now squeezed out of the ecosystem by the higher virality of the silly apps. This will allow engaging and useful apps to flourish in the subset of Facebook users whom Facebook clearly values the most, who are far more valuable for monetization, without spoiling the fun the giggling girls are having over in the other corner of Facebook.

It is in the apps’ interests to be classified correctly, therefore self-classification will be sufficient. The division will reduce Facebook’s need to police the apps, because serious apps will treat serious users more in line with their expectations and giggly users are more tolerant of highly viral tactics.

The 75% and 25% numbers are my approximations, based on polls about the Beacon program and forced invites, on the demographics, review of statistics, and a great dose of guess. The dichotomy is not firm, and the numbers may not be exactly 75/25. There is, however, a giggly majority and a serious minority, there’s greater conversion to inviting among the giggly users, there’s Facebook’s desire to be a social utility and thus to appeal to the serious minority, and there’s the problem of higher virality of the silly apps on Facebook, combined with the desire to have serious apps.

There are two ways in which Facebook would enable engaging useful apps:

change the distribution model from viral to directory and
segregate the users into groups.

Directory-based distribution would be bad news as it would replace competition with arbitrary choice, reducing the overall quality of apps. I believe Facebook understands this, as they have resisted this route thus far.

Segregation of users and apps into groups is the next natural choice, and I do not believe it has been explored. The minimum useful number of groups is two, and two groups might well be sufficient.

Facebook already has good data that separates the giggly users from the serious ones. I expect that serious users have been far more likely to change their privacy settings from the defaults. Why not start from there?

Update, July 24, 2008: Facebook has made a choice. Apps will no longer compete on a level playing field. Instead, Facebook will separate them into three tiers of preference. The replacement of competition with arbitrary choice by Facebook employees will obviously lower the overall quality, except as perceived by the particular employees making the choice. Yet it’s Facebook’s platform and their choice how to run it.

Who Should I Follow? Twitter Friend Recommendations

Fri, 13 Jun 2008 20:24:00 -0700

Who Should I Follow? Twitter Friend Recommendations

Yahoo-Google search deal is ugly news for startups

Fri, 13 Jun 2008 19:24:26 -0700

Google targets AdSense revenue sharing so that publisher gets a bit more than the next best option, which is usually Yahoo.

Yahoo running Google ads puts Google even closer to monopoly in contextual ads, reducing the revenues of all AdSense publishers.

Consumer startups struggling for a better model often rely on AdSense for revenue.

The alternative to the Google deal for Yahoo was, of course, accepting the Microsoft offer. Had they done so, these parties would be better off:

Publishers: with better AdSense alternative, AdSense would pay more
Consumer startups in the same ecosystem as startups publishing AdSense
Advertisers
Yahoo shareholders
Yahoo employees and middle management

While these parties would be worse off:

Google (getting a real competitor)
Microsoft (likely paying too much)
Jerry Yang

How can a public company CEO act in self-interest so obviously?

The silver lining for startups is the preservation of Yahoo as a possible acquirer. Yahoo shareholders don’t get a silver lining.

NVCA Model Venture Capital Financing Documents

Fri, 13 Jun 2008 02:12:08 -0700

NVCA Model Venture Capital Financing Documents:

When a larger than usual set of lawyers develops universal VC funding documents, they end up with documents that appear to satisfy no-one. This is too bad, because the idea of standard financing documents is great. I wonder if you could make each document fit on one page.

Redesign A First Step In Bringing Order To The MySpace Chaos

Thu, 12 Jun 2008 21:24:01 -0700

Redesign A First Step In Bringing Order To The MySpace Chaos:

MySpace is redesigning. 25% of all Americans visit MySpace monthly, making it largest site in the U.S. Disturbingly, 12% of all online minutes are spent on MySpace.

Garry Kasparov Griefed by Flying Penis

Thu, 12 Jun 2008 20:50:02 -0700

Garry Kasparov Griefed by Flying Penis:

In Soviet Russia, Second Life plays the world chess champion. RC helicopter with a giant penis attached buzzed around Kasparov’s opposition assembly, in real life.

Yahoo Announces Non-Exclusive Search Agreement With Google

Thu, 12 Jun 2008 15:58:02 -0700

Yahoo Announces Non-Exclusive Search Agreement With Google:

Startups are supposed to have better revenue models than AdSense. Yet Yahoo is doing just that.

Vobile, video fingerprinting tech (VideoDNA)

Thu, 12 Jun 2008 15:11:02 -0700

Vobile, video fingerprinting tech (VideoDNA):

Doesn’t modify the video. I infer from marketing materials the fingerprint is roughly 1% in size of the original. Doesn’t mention computational costs.

The less people know about climate science, the more they worry about global warming

Thu, 12 Jun 2008 14:24:03 -0700

The less people know about climate science, the more they worry about global warming

Judge Rules That Police Can Bar High I.Q. Scores

Thu, 12 Jun 2008 00:37:04 -0700

Judge Rules That Police Can Bar High I.Q. Scores:

Police in New London, CT, deny job interviews to people with IQ scores too high. Unfortunately, the article does not specify the threshold.

TinEye

Wed, 11 Jun 2008 19:11:05 -0700

TinEye:

“Give [TinEye] an image and it will tell you where the image appears on the web.”

Glassdoor.com - Salaries

Wed, 11 Jun 2008 18:50:14 -0700

Glassdoor.com - Salaries