Giovanni Bajo's swapfile

jaredsinclair:

I dislike iOS 7 so strongly that I feel inclined to begin this post with a disclaimer about how much I admire Apple. Apple is my hero. They’ve always inspired me to be better at what I do, even when I was an ICU nurse. But they are not perfect. I can and should criticize their worst work when I…

I tend to agree on most of this post. I really hate the borderless buttons in iOS7 and the Contacts app is really awful.

While you shouldn’t usually ignore a 30% market, if I were to start developing an Android application right now, I’d have no doubt: 4.0-support only. Let’s capitalize on the huge Android market share and declare that 70% of it is more than enough.

The problem here is not really saving on the huge amount of additional testing (most of it would have to be done on cheap/slow devices), but that you really don’t want to be forced to be limited to API level 10, and at the same time be forced to also develop a fully-compliant Holo interface.

Android [up to 4.3] is using the combination of horribly broken RC4 and MD5 as the first default cipher on all SSL connections. This impacts all apps that did not care enough to change the list of enabled ciphers (i.e. almost all existing apps).

Picture this. There is a big building, and on the top floor there’s a beautiful office where Sundar Pichai works. On the first floor, the Android team ships an operating system with very questionable security choices (of which this article is just the latest news) and no way to ship security patches in a decent timeframe to users; on the second floor, the Chrome team not only revolutionized the approaching of keeping software up-to-date in the desktop world (remember geeks screaming that it was unacceptable that it updated without user consent?), but is also spearheading many important efforts to improve TLS security in the real world.

I mentioned this before, but I’m really crossing my fingers that the brain merger between Chrome and Android goes right.

Apple and EA are both categorically denying a report that surfaced late yesterday that money exchanged hands to keep Popcap hit Plants Vs. Zombies 2 an exclusive to iOS. Reached for comment on the matter, both EA and Apple told us that “it’s simply not true.”

This story generated lots of fuss but I’m not sure what it is this about.

Samsung has publicly run a program in which they paid iOS (and non-iOS) developers to develop exclusive apps, only available on their Samsung Store, and/or shipped at launch as first-party apps for their smartphones / phablets / devices. They have been doing this for 2/3 years now. I didn’t see the whole tech world going frenzy about it.

In fact, even if Apple paid EA for this, I wouldn’t see what the problem is.

The linked article is a very interesting essay about a software developer that increased revenue of their app. Go reading it and come back, I’ll wait.

The focus of that article is on how adding a button “buy all” (to buy 10 different items that had to previously be bought separately) had a very positive effect on the revenue. This might be interesting, but I think it’s intuitive: an increase in convenience is surely irresistible, especially to the lazy consumer market.

But I would like to focus on the first part of the article, where they say:

Right around Christmas we […] increased the price from $1.99 to $2.99. We saw a slight decrease in purchases per user, but we saw a big jump in our revenue.

There is also a nice graph of the jump in revenue, and it’s impressive. But I’m always amazed by the sheer number of people (especially among geeks) that find this effect surprising.

The main reason is, I believe, the wrong belief that increasing the customer/user base is the single most important thing. Now, it is surely important to have a large number of customers, and the more the better. Since raising the price undoubtedly lowers the number of customers, the intuition would say that this might be a bad thing, since there is an exponential distribution between users and price they want to pay.

But what you really need to maximize is not the number of users, but the result of the multiplication between price and users. There is absolutely no certainty that raising the price lowers the users to the point where the result of the multiplication is lower. It depends on the kind of exponential, and that’s where the trick is.

In fact, there is one specific case where you can take for granted that you can raise the price and increase the revenue: when an item is very cheap (and in-app purchases in software are always cheap nowadays, speaking in absolute terms), such a price change makes basically no difference for someone who really wants to buy the item, but it produces a big effect on the revenue.

Another important paragraph is this one:

Throughout the first year we realized 3% of our customers were purchasing the first book. Of those 3%, 70% would purchase both books.

Which meant that of our paying customers 70% percent were purchasing everything we had to offer.

We were losing sales from not having enough content.

This is a smart realization. When you segment the market among customers, you should expect them to follow an exponential distribution. If they don’t, there is something wrong with your price.

In this case, you can think of this app being a software or a service sold in 3 versions: a free plan, a basic plan ($3 = 1 in-app purchase) and a premium plan ($6 = 2 in-app purchases). The free plan will be the most popular, but you should expect the basic plan to be far more popular than the premium plan. If 70% of your paying customers are purchasing your most expensive offer, either it’s not expensive enough, or you want add more tiers on the top, to further segment the market, and extract more value from customers that are ready to pay more for your product.

Xiaomi [….] innovative strategy: making high-spec smartphones at low-spec smartphone prices.

That’s really innovative, how I couldn’t think of it before. Let’s produce Ferrari clones at Tata’s price, I’m sure they will be a huge hit, people will love them.

Google hasn’t provided details on its new encryption efforts but did say it would be “end-to-end,” meaning that all servers and fiber-optic lines involved in delivering information will be encrypted.

Engineers at Google seem to believe NSA is tapping infrastructure fibers that connect their data centers, and that do not always transmit encrypted information (they are probably MPLS level-2 VPNs, which are unencrypted on the basis that they are not part of the public Internet, but routed directly by the ISP — and you could trust your ISP before Snowden came out).

But the point is NSA is probably just subpoening their TLS private keys, and Larry Page is just forced to lie to them and in public.

This doesn’t mean that encrypting inter data-center connections is wrong, on the contrary; it’s just not the most useful anti-NSA measure we cant think of right now. You just need to admit that Google is probably complying to NSA subpoenas and gag orders, and act accordingly.

Fortunately, Google has already deployed TLS perfect forward secrecy, which is the correct crypto solution to the fear of TLS private keys being compromised.