[Cogito]

It's easy to presume that sunk costs should have no weight when considering the relative value of different options.

For many endeavours a certain amount of investment is required before returns are seen - investments in time and money in particular. For this reason there is almost always a cost to be considered switching between endeavours.

So, while I agree that throwing good money after bad is a silly thing to do, the consideration should be how much more investment is needed to make this endeavour profitable vs how much some other endeavour will take to become profitable AND how much it will cost to switch.

You shouldn't weight how much has already been spent in this consideration, however you have to understand the impact existing investment has had on the road to profitability.

Time and money already spent is a crude metric that lends itself to the Sunk Cost Fallacy, but considering the metric is not the same thing as falling victim to the fallacy.

[EDIT] The following comment explains this really well through an example:

https://news.ycombinator.com/item?id=5722032

[raylu]

I think you misunderstand sunk cost. Sunk cost basically just says to always employ a purely-forward-looking strategy when evaluating costs and benefits.

> vs how much some other endeavour will take to become profitable AND how much it will cost to switch.

Wait, what are these two costs? Aren't they the same? Where did the cost of switching come from?

[Cogito]

> I think you misunderstand sunk cost. Sunk cost basically just says to always employ a purely-forward-looking strategy when evaluating costs and benefits

I don't think I misunderstand sunk cost, but perhaps I didn't explain my point well enough.

I agree that you need to be forward looking, but whilst looking forward it can be important to consider existing investment. Existing investment is not ALWAYS worth nothing, so its value needs to be accounted for. Existing investment is also a metric that, whilst lending itself to entwining one in the Sunk Cost Fallacy, is an important part of answering the question "how much more investment is needed?"

The sunk cost fallacy applies in a situation where previous investment has been lost. The typical example is in gambling, where each spin of the wheel is independent of how much you have already lost on the table. When in a startup, however, previous investment is not necessarily lost, so switching to a new one is (typically) not the same as continuing one you have already invested in.

>> vs how much some other endeavour will take to become profitable AND how much it will cost to switch.

> Wait, what are these two costs? Aren't they the same? Where did the cost of switching come from?

The cost of switching is a way of characterising things like lost earnings and payouts that are not directly related to starting a new endeavour. They are typically, though not always, constant regardless of the new endeavour being considered, and so it makes sense to keep them separate.

Examples of switching costs:

* paying out a phone contract;

* loss of long service benefits (if someone is close to receiving long service benefits they will typically wait to receive them before switching jobs);

* non-compete clauses.

[pyre]

  | For many endeavours a certain amount of
  | investment is required before returns are seen

Compulsive gamblers tell themselves the same things. What is really needed is a objective view of the situation.

[Cogito]

Are you disagreeing that investment is often needed in order to gain profit?

I definitely did not advocate gambling money away, but rather that "an objective view of the situation" requires you "to understand the impact existing investment has had on the road to profitability."

In a follow up to another comment in this thread, I specifically call out the difference between gambling, where (usually) any investment so far has no impact on future earnings; and investing (in particular investing in startups), where the time and money invested so far can have an impact on future earnings.

To be concrete, the next spin of the roulette table has no correlation to the amount of money you won or lost on the previous spin, but your chance of selling an app on the app store in the next month is directly correlated to how much of the app has been written already.