20 June 2014

On dating

What affects the divorce rate? Two things come to mind: the cost of divorce and match stability. Both nominal (legal fees, etc) and social costs of divorce would affect one's decision to get divorced. Arguably, lower cost is desirable. In general, we would not want people to stay married when both parties wish to leave. But why do people want to get divorced in the first place if they married under mutual agreement? There is probably some sort of information problem. For one, it is difficult to know and understand a person, especially when the person may purposefully act differently to appear more desirable.

Match may break down not only due to violation of individual rationality, but also due to a blocking pair. That is, there is problem of "cheating". Unlike the lower cost of divorce, match instability is quite certainly undesirable. It is costly to get married and then get divorced. More importantly, however, divorce after having children imposes negative externalities (of course, children themselves increase the cost of divorce so the externality is at least partially internalized; the problem isn't so much the act of divorce itself but the effect of poor parental relationships on the children).

Currently in the US and probably in many other countries, the conventional procedure to address this information gap problem is dating. People go through the process of dating to learn about the other person and to determine whether the person is acceptable as a marriage partner.

The problem with dating is that it addresses (at least partly) the individual rationality problem, but it fails to prevent a blocking pair simply because the convention dictates that you date one person at a time. Suppose A and B are dating and C comes along and shows interest in A. A suspects that C might be a better partner than B but of course is not sure. A must make a decision to either break up with B in order to start dating with C or stay with B. Of course, if A decides to break up with B but ends up not liking C, then A's chance to get back with B is slim. Thus the option of dating with (or put differently, learning about) C is highly risky. Thus the current convention discourages A-C match to form even if it blocks A-B.

Hence, rather ironically, the society's harsh punishment toward cheating leads to less stable marriages by preventing agents to explore their options. Since break up is much more costly after the marriage, one has more incentive to "explore" the options after the marriage (if you get caught cheating while dating, break up is certain; if you get caught cheating while married, divorce may or may not happen). Even more ironically, the society likely punishes infidelity to maintain the family structure, in which case "cheating" before marriage would be irrelevant.

In short, the fact that dating with multiple people simultaneously is a taboo likely result in less stable matches (in terms of marriage) which in turn results in more divorces. Since divorce puts higher social cost (notably due to externalities imposed on children) compared to break-ups before marriages, the society would benefit by shifting the equilibrium to one in which fidelity to a single partner while dating isn't asked for.

14 June 2014

Improving browser navigation

TLDR: Let the browsing history be carried over when a link is opened in new tab, so the user can click the back button to, well, go back.

The main navigation tool of any browsers is a back button. My Chrome, for example, only has back, forward, and refresh buttons. And this is for a good reason. An HTML document may contain a hyperlink to another page, but the linked page may not have a hyperlink to the original document. The browser's back button is essentially the only way to go back to the parent page from a child page.

All the modern browsers I know have introduced a major UI change in early 2000's: tabs. Tabs have dramatically lowered the inconvenience of having multiple pages open at the same time; navigating between different tabs is much easier than between windows. Hence, a viable (and probably popular, although I do not have data for this) approach to navigation is following a link by opening a new tab rather than in situ.

This style of navigation in some sense makes back button obsolete. Once you open a child page in a new tab, you can go back to the parent page by clicking the original tab. This is also convenient when you want to visit the child page back again. In particular, if one wants to go back to one of multiple child pages, using tabs is far more convenient than navigating through back and forward buttons.

One major draw back of using tabs to navigate, however, is that one loses track of parent-child relationships. A new tab page has no history; if you open a link in new tab, you cannot go back to the parent page through the back button. The child becomes oblivious of its parent; all tabs are independent of each other.

Suppose I am doing an online research. I open up a search engine (page A) and find few relevant links. I open those pages (B1, B2, B3) in new tabs. Pages B1 and B2 contain links for further readings that may be useful. So from page B1 I open pages C1 and C2 and from page B2 I open D1 and D2, all in new tabs, since I might come back to the old pages later. I finish reading D2, and now I am interested in other types of further readings recommended by B2. So I want to go back to the parent page, but now I am not sure which of 7 other tabs that is.

This type of navigation problem usually gets resolved in trial and error fashion which is obviously not very convenient. Only if that back button on page D2 wasn't grayed out. But wait, it doesn't have to be. There is a simple solution. When a child page is opened in new tab, the parent page can pass its history to the child tab.

When a user reaches a page through a sequence of hyperlinks, the page clearly has a history that might be often relevant for the user. Why should that history be lost just because a page is opened in a new tab?

I am aware that there are browsers/extensions that color the tabs to group the family of pages. This is sometimes helpful, but still it does not give any information on vertical hierarchy (so it would not have helped in the example above). The colored tabs can also be often visually jarring and hence confusing. Simply carrying the history over, on the other hand, is much more natural. There will be no visual changes and the back button will always behave the way the user expects it to.

I doubt that carrying browsing history over to tabs will cause much confusion since it simply adds the ability to "go back" when a link is opened in new tab. I am quite certain that in many occasions it will improve the navigation. So why don't browsers implement it?