HN user

abrenzel

157 karma
Posts5
Comments44
View on HN

This article is sensationalism dressed up as economic reasoning. In the opening paragraph, the author fails to make the distinction between profits trending toward zero versus prices trending toward zero in a free market. This undermines his argument even without the very real and substantial caveats he waves away later in the piece.

Second, while the marginal cost to sell one copy of a game is effectively zero in electronic marketplaces, the gross cost of making a game is never zero, not even for indie developers, because if nothing else it costs them their time. What we should then expect to see is a situation where, assuming a game is successful, a game sells at some non-zero price the market will bear until it earns its costs back, with steep discounts following. Of course, it is more than possible the game will simply lose money.

We should expect this to be true even in a world where games are sold purely in electronic form and those markets are flooded with free content (which is to say, developers willing to sell their games at a loss). Come to think of it, that's pretty much precisely what we observe in the games industry today. Where's the beef?

Yes, I contradicted your assertions that

a) it is MIT and JSTOR that have primary discretion in whether a criminal case moves forward (it is the government's sole discretion)

b) MIT and JSTOR are in fact responsible for bringing criminal charges (they are not - they can only bring civil charges)

c) that the prosecutor is responsible for the severity of the penalty and/or the defendant's emotional response to that penalty

My disagreement with your third assertion was more implicit so let me clarify a bit.

First, it is not the prosecutor's job to question whether a law's penalties are in proportion to the crime it proscribes when deciding whether to bring a case. Discretion over the severity of the punishment is left to the sentencing phase of the trial if the defendant is convicted, and it is highly likely that Aaron's sentence would not have been the maximum had he been found guilty (a fact I am sure his lawyers made him aware of).

Second, under what reasonable standard can a prosecutor be held personally responsible for the emotional state of the defendant? Should it be acceptable for criminal defendants to pressure prosecutors into dropping cases by threatening self-harm or suicide in the hopes that a public outcry will harm the prosecutor's career? Try to ignore for a moment that the defendant in this case has your sympathies. Would you accept that tactic from a serial killer or rapist?

It may be that the law itself is unnecessary or counterproductive. I'm certainly open to the argument that at least publicly funded research ought to be open to the public. Yet it is still the law of the land. From the facts of the case, Aaron committed an obvious crime and behaved as though he knew it were a crime. The potential price of civil disobedience is that you will in fact end up punished for it. In the end, his story (like Rosa Parks and others before him) may end up bringing about the change he wants. But to say that the prosecutor abused her authority or was personally responsible for his death is an emotional response without basis and runs counter to the very idea of a criminal justice system.

Strictly speaking this is not correct. JSTOR and MIT cannot bring criminal charges; only the US government may. They can bring civil charges on their own behalf in addition to the criminal charges, and whether the 2 parties agree to settle out of court is relevant to any civil charges brought against the defendant. This is the basis for the difference between civil and criminal law - civil law covers disputes between 2 parties and serves as a framework if the 2 parties cannot reconcile.

On the other hand, criminal actions are not just about the grievances between the defendant and the victim of the crime. In principle, crimes are committed against the people themselves (hence the case naming scheme exemplified by the United States vs. X or The People of Illinois vs. X). Even if the victim isn't particularly interested in pursuing the prosecution, the prosecutor is still within his rights to try the case. Indeed, in many violent crimes such as rape, the victim is not interested in aiding the trial. While this can sometimes derail the prosecution, it need not. Pursuing the case absent the victim's full cooperation cannot ipso facto be considered prosecutorial abuse.

I just see this as an incremental development, rather than a revolutionary one. As I mentioned above, the server is still very involved with UI events and navigation. If the user causes a UI event that saves data to the server, even if there is client-side validation it will often be the server that responds with an error and forces the page to do something with that. The nice thing is that we now get to encapsulate the server's influence on the application into definable actions, rather than by a single request. That's a rather good thing for us developers, but it is still well within the bounds of the traditional HTTP model.

As for Coffeescript, anything that tames some of the nasty bits of Javascript is fine by me. I don't dispute Coffeescript's usefulness. My point was just that Coffeescript is like syntactic sugar for Javascript, not a paradigm shift in client side coding.

Well yes, now you can perform more manipulation of the page without needing a roundtrip to the server, but saving the state of the application still involves one, unless you are the type of radical early adopter who is going to try to do it all with HTML5 local storage.

The relative balance of the work may have changed, but I don't see it as a revolution except in terms of user experience, where AJAX has definitely drastically changed the way users perceive a web page.

Otherwise, the development model of HTTP - the server sends me something, I send a reply, and it sends me back something else - is still alive and well. That's to be expected, since browsers were built with HTTP in mind and the supporting technologies, Javascript included, all must live within that box. The server is still intimately involved with the behavior of the page, but its involvement is encapsulated into individual actions rather than an entire page load (i.e. the server returning an error can be responded to in the context of the HTML form that caused it, rather than with a page redirect or some such). Again, I don't deny that's a helpful thing.

Some things from this article just don't make sense to me.

Like, why is AJAX/client-side MVC pointed out as something that overcomes the HTTP request/response cycle? No, it doesn't. It just means you don't have to refresh the entire screen every time a request is made. Whether the server now sends back "a wad of HTML/javascript" or some JSON to be parsed by the Javascript MVC framework du jour, nothing you are doing is transcending the HTTP protocol either way. AJAX changed the way users interact with web applications, but it did not change anything really fundamental about their architecture.

Second, Coffeescript does not liberate you from Javascript. It IS Javascript. It makes it easier to control some of Javascript's difficult areas, like the meaning of "this" as execution context changes as one example, but it doesn't suddenly give you license to write browser code the way you would write Python or C++.

In any case, let's not confuse convenience with paradigm shifting.

Greece defaults 15 years ago

Absolutely. Indeed risk is not even just about the "customer's" (in this case, the nation of Greece) inability to pay, but also counterparty risk. It's not just Deutschebank that's on the hook for defaulted Greek debt, but every counterparty who wrote them credit default swaps (CDS) on that debt. That's what triggered the financial crisis in 2008 - as Bear Stearns' and Lehman Brothers' cash flow from their debt holdings dried up, redemption calls for their CDS metastasized though the financial system.

If anyone is wondering why the core Euro-area (not to mention the US government) is so concerned about Greek (and Irish and Portuguese and Spanish and Italian) debt, it's less about some abstract political commitment to EU unity than it is the simple fact that a true default will destroy the European banking system.

Throwing cheap cash at the problem will work until the day that it suddenly doesn't. While the proximate cause to the original financial crisis was cash flow, that is not the ultimate cause for these problems. The problem is that risk in the financial system is (still!) extremely opaque, and it is not so much that banks are illiquid as there is almost no circumstance in which they could procure enough cash to meet calls on their outstanding CDS in the event of some "unexpected" event like subprime mortages or Greece going into default.

The US government learned the hard way what would happen if they allowed one of these overlevered banks to go under as they did with Lehman Brothers in 2008. On the other hand, national governments do not have enough capital to possibly cover the total liabilities in the financial system, nor can they even predict when and where that capital might be needed.

As with many things in life, there is no real good solution to this, and there is still lots of pain ahead.

From the looks of it, the program merely learned the rules of the game by doing textual analysis of the manual, and maybe got a few strategic hints as well. As for actually learning to play the game _well_, my sense from the article is that the program then used more conventional machine learning techniques to test and adopt winning moves.

I've been working on a new site in Pyramid (the successor to Pylons), and I have to say so far it's been an impressive experience. They took a bit of a different philosophy than Django and Rails. In those 2, you can customize some things, but there are also default options, and particularly things like the ORM are built into the framework. Pyramid behaves more like an application widget where the pieces of a modern web app (ORM, security, templating, URL routing) can be fit into its slots.

I did enjoy the article's touching on Django. It looks like it has gotten less monolithic than in prior versions. As I recall several years ago, Django really was like the Rails of Python, with tons of default options but some difficulties in customization. It looks like they have moved away from that somewhat.

Not just a housing bubble but a bubble in the exact same industry sector as well!

Of course, a bubble is fundamentally a gross mismatch between future expectations and actual returns. That's why there's no learning from the past, and why it's always "different this time."

More importantly, bubbles don't occur in a vacuum. The 1990's tech bubble wasn't just because the internet was new, and everyone thought it was going to be great. The 1990's also featured low interest rates, the repeal of Glass-Steagall, and other loose policies conducive to asset bubbles. Sound familiar?

Such policies force money to chase yield, which means risk. As it does so, the market sends false signals about the true value of its companies, which reinforce themselves until some "Black Swan" event reveals the imbalances. In 2008, that event was the fall of Lehman Brothers.

As I noted below, if the national government continues to tighten its fiscal and monetary policy, then this bubble will be over before it really began. If, however, the government reloosens after the end of QE 2, it will most likely be full steam ahead.

That was a good point in the presentation I thought, but I objected on the slide that suggested you wouldn't learn anything from it either (slide 28). How could you not learn something from the experience, even if you fail hardcore?

I think a lot of conventional analysis of investments is looking more foolish than usual because the market isn't keying off what people think it is. Three major things have dominated market movements since the 2008 crisis:

1) Rock-bottom interest rates (and a depreciating dollar) 2) Quantitative easing 3) Enormous deficit spending

All of these things, through mechanisms both simple and complex, have driven money into equities and speculative investments such as start-ups.

While interest rates will remain low for quite a long time and deficits will no doubt continue, QE2 expires at the end of the month. Additionally, the proportion of the deficit is moving away from stimulus into structural spending such as Medicare. The natural response by the major market players is profit taking. Hence, the IPOs at bubble-like valuations (LinkedIn, Pandora, etc) are running into the teeth of a "smart money" evacuation of the stock market.

If the Fed and government overcome political pressure against further loose policy (always possible if the economy continues to weaken), then I imagine the LNKDs and Ps of the world will pop back up to the eye-watering valuations they attained at the time of their IPOs.

If national fiscal/monetary policy continues to tighten, then the bubble will have been ended before it really began. In which case, LNKD and P will continue to decline, and I would not be surprised to see some of the other hot properties like Zynga delay their market entries. The market for VC funding will also become substantially tighter.

Am I a loser because I didn't feel like downloading a Scala compiler or translating this to a language I do have an interpreter for, so I just got the table from man ascii, counted the characters in the string, and then did the transformation by hand?

I thought this was going to be another one of those benchmarking articles, but it was really a great exercise in how lateral thinking can produce clean solutions.

The more I hear about the Lisp family of languages, the more I want to give them a try sometime...

Poll: Tech Bubble? 15 years ago

Is this a full-on bubble? No, it's probably nothing like the 1990's. But I still think the truth is that today's monetary situation is creating an incentive to chase yields, and it is not out of the question that this is showing up in VC valuations just as it is showing up in commodities.

The fact that these valuations seem to be stemming more from the money others have put in (or said they would be willing to put in) rather than the revenue of the company itself ought to arouse deep suspicion. Yes, I understand part of a valuation is examining the pricing signals being sent by the market, in this case VCs, for the particular company. Still, look at Twitter. Some valuations for it have come in at $10 billion when it's been admitted their yearly revenue is $150 million. If Twitter were to IPO at that valuation, that means 99% (+/- for whatever other assets they might have) of their float would be pure leverage. For many other tech start ups, the same sort of situation exists even if the magnitude is not as great.

For those down the line in the thread talking about the better development tools/programming languages/hardware in relation to this topic, let me ask: how do better development tools justify higher valuations? Certainly, I'd rather be developing a web app with Python/Django today than in ASP in 1995, and running it on today's server hardware with today's replication and load balancing strategies. There's no doubt the development cycle for many kinds of applications has been radically shortened. Yet, the revenue model for many of these internet companies is only slightly clearer than it was in 1998. That is not true of every company, only some; but it is true of enough that I think it is cause for concern.

Experimentation is a wonderful thing. It is the only way the economy advances and people's lives become better. Part of the "startup industry"'s job is to provide that experimentation, and by its very nature most will fail. The problem is when the investment community begins to pay a large speculative premium for that. Even if the resulting "bubble" is not an economy-destroying one as it was in 2000 or 2008, it can still do a great deal of harm.

Poll: Tech Bubble? 15 years ago

But don't you think we're seeing those in commodities and many stocks, including but not limited to tech stocks? You wouldn't see it in real estate, because popped bubbles don't reinflate. Whenever I see a company like Cisco drop huge %'s overnight not on a loss, but on "somewhat worse than expected" revenue, my alarm bells start going off.

Eventually, that is the beneficial effect of a bubble bursting. On the downside, the short term of a bubble bursting brings risk aversion and particularly aversion to debt.

I'm in agreement with the linked article that the "valuations" of companies like Facebook and Groupon are facially absurd. Facebook is worth $60 billion? On what planet? Yes, they have a lot of users, but are they monetizing them to the degree necessary to sustain that valuation? Facebook's books are a black box so we may never know, but my strong suspicion is no.

The little mini-bubble going on in some tech startups as well as the bubbles building in asset classes like commodities are the direct result of nearly every nation in the world pursuing the "beggar thy neighbor" (read: currency devaluation) strategy at the same time. The money pouring off the printing presses and/or being borrowed by national governments is heading straight into these types of investments. When that cycle ends, prices are going to drop like a rock.

Back to the original comment, the "boring" or smaller companies being ignored in the frenzy will benefit from that, eventually. But first the damage done by bubble will have to be healed, and that can take awhile.

I am intrigued by Watson. NLP is definitely something that has failed to live up to its billing so far. I will be very interested to see what kinds of questions Watson is good at and which he tends to miss, to see if there are any patterns there.

Going forward, the other real questions will be: is Watson overfitted to the problem of solving Jeopardy questions, and how practical is the technology? The former is a real risk to the general applicability of Watson's technologies, the latter is a question of who can afford it. The article mentioned on commodity hardware, Watson takes about 2 hours per clue. They only achieve reasonable response times by using about 3000 cores. That limits the potential audience.

Either way, I'm very interested to see what happens next week. I watched the demo videos on YouTube and it was quite cool.

But the interpreting process often is a form of compilation - e.g. many interpreters compile source to bytecode and interpret the bytecode - to say nothing of JIT, which after all stands for just-in-time compilation!

Compiler design I think will always be a vital field. The LLVM project I think shows the direction we are headed - with more and more detailed bytecode to allow for better optimization before a 2nd pass compiles down to machine code.

Loved the comment about it being a "switch." This is exactly what I found when I went to Turkey after 2 years of studying Turkish. I was in a program where we had a 4 hour class in the morning taught about 80% in Turkish. The first week or two was terrible. Then the "switch" flipped. I wasn't a fluent speaker and still missed words and phrases while listening, but all of a sudden I could follow the class, do the lessons correctly, and ask semi-intelligent questions. It's a great moment when you're learning a language.

Groupon was silly not to accept Google's $6bn offer. The profit margin that justified that valuation is sure to drop over time, as their business model is easily replicable. Does that mean Groupon or that business model is going to go away? Of course not. It does mean, however, that $6bn was probably the high-water mark for the valuation a company like this can achieve.

The biggest threat is not from the consumer side - who wouldn't love 50-75% off deals? - but from the business side. Rumor has it that business are losing anywhere from 75-90% on these coupons. Businesses will accept a loss leader now and again, but there are two threats here:

1) Competitors will offer businesses better deals by taking less as a middleman fee, hurting Groupon's profit margin

2) There will be few repeat customers on the business side, as Groupon purchasers don't actually return to the businesses based on the initial discounted deal

Both threats are serious.

It may be more of an aesthetic preference, but the pricing signal makes more sense to me on a per-service basis than it does on a tiered internet service basis. Tiered bandwidth plans are inexact - should I buy the 250 MB plan or the 5 GB - and require a lot of thought as to what you will actually use. If the charge is per-service, you know exactly what you're buying with each additional service you consume. That's my thinking anyway.

Well you're right in a sense that Netflix probably pays the appropriate amount purely for what's outbound from their servers, but that last cost does count and the cost will be borne by someone - either by what Netflix pays the ISP for its data in particular, or by the users that use it. The problem with Net Neutrality is that you're basically saying "everyone should pay for what some content providers provide to some users." Either ISPs "discriminate" against the content provider or the end users, or everyone ends up paying more. Currently IMO, the best world would be to allow the ISPs to charge Netflix properly for the total cost of providing it, which would build in to Netflix's subscription price and properly reflect to end users the true cost of Netflix's services.