3 ms·

I'm sure it's just a level of optimization that's hard to solve because of a few reasons. a) it's a fairly hard technology problem. Large container ships can ca

I'm sure it's just a level of optimization that's hard to solve because of a few reasons. a) it's a fairly hard technology problem. Large container ships can carry 20k containers. If each of those has a spot, and you have to optimize, the search space is suddenly looking like 20,000!, which is just a wildly big number. So assuming you have some sort of technological solution and know where every container should go, b) you need to move all the logistics around so that containers arrive in the correct order to get loaded in the correct spots. You need to do all this just to massively increase the complexity of your billing model to offer slightly different rates of insurance. I'm sure there are teams looking at ways to make this better (just stipulate in the contract that any super valuable containers arrive early so they're loaded lower), but this is a HARD problem to solve completely.

OldHand2018 6y ago
I'm not an expert on ocean shipping by any means, but I would assume that cargo value is such a low priority when it comes to determine loading strategies that you would only need to optimize around a local area of maybe 100 containers. Which would not make it a hard problem at all but it could be pretty pointless depending on how many containers go overboard in a typical "lost container" event. Also, consider that from the point of view of the people loading the ship, it's all insured. And if you didn't get the insurance then you understand the risk. Why would they care? They aren't the insurance company. What's their incentive?
ClumsyPilot 6y ago
This is such a HN thread.. The same effort is better invested into making sure you are not losing containers. It's like saying 'we can optimise policing so that stabbings only happen to the right people'
- GauntletWizard 6y ago
  Do you think that doesn’t happen? Police departments absolutely prioritize certain areas over others - Places where crime is unlikely to happen but should be dealt with swiftly, places where crime is very likely to happen and they have resources on-hand. The net losers tend to be “The Average Person”, for whom police response times in event of a burglary or domestic incident is measured in tens and fifteens of minutes rather than fives. Even “Even spread” would be a serious topic of debate - Even spread over what? Geographic area? Residency? Where people are likely to be at a given time? Those are three radically different answers in any commuter city.
eru 6y ago
A big search space is not a detriment to optimization. It just keeps you from reaching provable optimality, but doesn't keep you from saving on costs compared to doing nothing. Same for the rest of what you describe: yes, a perfect system would be a lot of hassle, but making any effort at all is sure to yield benefits. You don't need to change the billing model necessarily. The customers don't have to know that you are doing this kind of optimization. Of course, the more you can expose incentives to your customers, the more efficient you can get. But even if your containers arrive in arbitrary order and you have to make your decisions on where to place them 'online', ie before the next container arrives; as long as you have any choice left at all, you can try your hand at optimization techniques.

OldHand2018 6y ago

I'm not an expert on ocean shipping by any means, but I would assume that cargo value is such a low priority when it comes to determine loading strategies that you would only need to optimize around a local area of maybe 100 containers. Which would not make it a hard problem at all but it could be pretty pointless depending on how many containers go overboard in a typical "lost container" event. Also, consider that from the point of view of the people loading the ship, it's all insured. And if you didn't get the insurance then you understand the risk. Why would they care? They aren't the insurance company. What's their incentive?

ClumsyPilot 6y ago

This is such a HN thread.. The same effort is better invested into making sure you are not losing containers. It's like saying 'we can optimise policing so that stabbings only happen to the right people'

GauntletWizard 6y ago
Do you think that doesn’t happen? Police departments absolutely prioritize certain areas over others - Places where crime is unlikely to happen but should be dealt with swiftly, places where crime is very likely to happen and they have resources on-hand. The net losers tend to be “The Average Person”, for whom police response times in event of a burglary or domestic incident is measured in tens and fifteens of minutes rather than fives. Even “Even spread” would be a serious topic of debate - Even spread over what? Geographic area? Residency? Where people are likely to be at a given time? Those are three radically different answers in any commuter city.

eru 6y ago

A big search space is not a detriment to optimization. It just keeps you from reaching provable optimality, but doesn't keep you from saving on costs compared to doing nothing. Same for the rest of what you describe: yes, a perfect system would be a lot of hassle, but making any effort at all is sure to yield benefits. You don't need to change the billing model necessarily. The customers don't have to know that you are doing this kind of optimization. Of course, the more you can expose incentives to your customers, the more efficient you can get. But even if your containers arrive in arbitrary order and you have to make your decisions on where to place them 'online', ie before the next container arrives; as long as you have any choice left at all, you can try your hand at optimization techniques.