I have a miniature version of this, as I'm in charge of the duty schedule at my flying club and I wrote a program to generate the schedules for me.
There's a function which takes a schedule and generates a score. Leaving a spot unfilled or scheduling someone who isn't available subtracts lots of points. Scheduling someone much more or less often than the median subtracts points based on how much of a difference there is. Scheduling someone on a day they prefer adds points, and so forth.
Then it's a simple hill-climbing algorithm starting from an empty schedule. I get people's availability and preferences into the program, hit run, and out pops a schedule.
I learned early on that I must check the result before publishing it. The algorithm itself works fine. Hill climbing probably isn't finding the absolute optimal schedule, but it does a good job. The problem, of course, is the scoring function. Sometimes it turns out to be missing some important constraint. Much more often, I need to tinker with the weights assigned to various conditions.
People assign too much agency and intelligence to computers. They are, of course, just tools. A nail gun will put a nail into whatever you point it at, whether it's drywall or flooring or feet. Likewise, a computer optimization system will optimize whatever function you tell it to, and will probably do an excellent job, but it's up to you to make sure the function accurately describes what the humans want.
Back when I was enlisted, my sergeant had a bit of a conundrum. Troops were grumbling about being repeatedly assigned to the same posts, and we didn't have enough time during the shift to scramble the roster while ensuring adequate manning.
I wrote a simple Excel macro which randomly assigned security personnel to posts they held qualifications for. Since it didn't have any knowledge of personality conflicts or preferences of the troops getting assignments, the results generated by the Rosternator 3000 were generally hated by all involved.
In the end the tool was still useful - the mere threat of a randomized roster was enough to get everyone to quit grumbling.
How many personnel/posts were you assigning, out of curiosity? I've seen a similar problem, not automated yet, with around 20 staff members, who have varying conditionals which would have to be accounted for. It's not necessarily worth my time to automate for such a small # of people, then again it's a fun puzzle.
The current approach is a 2-month rotating schedule. Since there are day/evening/overnight shifts, with overnight being the least desirable and evening the most desirable, it was deemed most equitable to simply rotate members through different 2-month schedules. This way, everyone across the year will be doing the same sets/timings of shifts in different months. This becomes very important for evening shifts, in particular, since transitioning to/from that schedule requires time to overcome the jet-lag induced.
Roughly seventy personnel to about half as many posts. This was during the surge in Iraq, so half (or more) of our garrison was out in the sandbox at any given time anyway.
This was a shift-specific tool, so it only had to account for certifications (e.g. a Response Force Member wouldn't be put into a Controller/Dispatch position). From what I recall, the post listing itself was on one sheet with personnel/qualifications on another. Posts were iterated by priority, so critical manning positions were always appropriately staffed from the source roster.
I wonder if you could augment this further. Assign a scoring system for various conflicts and preferences. Start with the randomly assigned roster and swap random people as long as it improves the overall score.
The problem here is that these conflicts and preferences are constantly shifting. Two people can be good friends one week and decide they hate each others' guts the next week. And then they can go back to being friends again the week after. The Rosternator 3000 can't keep up with this so you're always going to be generating bad schedules for a certain portion of your people.
This is one of the reasons good retail managers ignore the advice of their time-keeping systems and give peoples' requests preference. The time-keeping system ignores all of the human aspects of the problem to focus on the least-important aspect of keeping a store running--making sure there are warm bodies in it.
>Two people can be good friends one week and decide they hate each others' guts the next week. And then they can go back to being friends again the week after.
So... you're saying soldiers are like girls in high school?
This thought did occur to me, but I didn't have the time to continue iteration past the first revision. Police blotters don't write themselves, as it turns out.
If I had to do it today, I'd probably put the list of pair scores on one sheet (e.g. Snuffy/Smith at +5, Smith/Jones at -3) and perform iterative swapping as you suggested.
Note that this is the "stable roommates problem" and does not necessarily have an "optimal" solution (where "optimal" means "no unmatched pair of people would've rather been matched together". Eg, if Albert prefers Bob over Charles, Bob prefers Charles over Albert, and Charles prefers Albert over Bob, and all three prefer being matched with the last person over anyone else).
Hi, I did my masters research into this very problem. A lot of research here does just simply revolve around iterations of hill climbing [1]
One thing which has been done quite a lot (and that's not to say you don't already do this), is to measure the gradient of improvement of your fitness function, and if it's been flat for too long, to relax the conditions/weights by X percent (X becomes another thing to tinker here, but I start with 10%). The flatness here suggests that you may be in a local optimum in the candidate space.
This will allow worse solutions for many more iteration, (and a graph tracking the fitness will go the wrong direction), but in the end, you might find that you end up with a solution that is much 'fitter' than before the algorithm relaxed the conditions. (In the academic papers I read at the time, various words were used for this, sometimes it was called 'reheat')
It seems that even if you don't do this, you know about it intuitively anyway since you talked about tinkering with the weights, just thought I'd chime in to mention that the tinkering part is often automated now.
That's cool. I'll keep that in mind if I need improvements or just feel like doing something more with it.
I don't think tinkering with the weights is quite the same thing. That happens when the algorithm is doing a good job of optimizing the scoring function, but the scoring function isn't doing a good job of representing what people want.
It does remind me of another thing I occasionally do when I don't quite get the results I want, which is to manually change one of the assignments and then have it climb the hill from there. Sometimes I'll just clear some assignments and re-run, which is pretty much a manual, ad-hoc version of what you describe.
Ah I understand, half the battle is terminology (I was introduced to the 'weights' as hard and soft constraints, with corresponding penalties against the fitness function score).
Even though I did research on it, I found that the field is subject to a lot of colloquialisms (for lack of a better term) when it comes to different research cliques, and therefore the terminology I use isn't any more of less correct as a result.
It seems like you've got as good a handle on it as anyone.
My research was mainly involving investigating parallelism of the above algorithm (not very friendly to parallelism without work).
Nowadays, there's so many academic open source optimizers, as well as commerical, that I would usually tell people if you're solving for a business case, it's often more cost effective to frame your problem in a format digestible by the solvers and let them at it than writing the solvers yourself.
Of course, where's the fun in that? I'm jealous that you got to use this for a problem that you had, rather than in my approach which was for a problem I didn't need solved myself. I bet that was satisfying.
Before trying Simulated Annealing, try out all the industry grade constraint optimization solvers! They’re surprisingly more effective on average case inputs. I had to learn this the hard way when, for a class project, I was thoroughly beat out by people who just used dedicated solvers for TSP.
I'm sure other approaches can be much more effective, my research was almost 8 years ago and was framed around trying to parallelise one of the algorithms.
So my focus was half on the theory behind these solution search algorithms and half on parallelism and the associated theory.
I've a gut feeling that a superior algorithm often gets you better ROI than parallelising an existing solution.
The TLDR of my findings was that for the algorithm in question, it wasn't greatly suited to parallelism. Performance didn't scale linearly with cores due to too much shared state, and the best you could hope for was seeding each parallel run of the algorithm with different randomisation seeds and it's own state and bank on certain runs not getting caught in local optimum.
So it got you better performance, but not so much that you'd jump right in to doing work to parallelise. Probably better to use algorithms that are built to be embarrassingly parallel from the ground up.
It is similar, but Hill Climbing moves one parameter at a time. While gradient descent moves along the steepest gradient adjusting multiple parameters at a time, but it requires the function to be differentiable
It's entirely possible that your particular case has constraints you can't easily express as a linear program. But if you can, go ahead and use the nuclear option!
Reading about scheduling reminded of a conversation I heard about involving an ER doctor.
In that ER center, the scheduling of ER Doctors was done manually, by one of the ER doctors. Of course not the most senior doctor. But not the most junior doctor either, cause that doctor will be bullied into giving someone a preferred schedule. So the job was given to one of the junior doctors, like 3rd or 4th newest doctor out of about 15 ER doctors.
These are doctors with big egos and with family, and all have legitimate reasons not to work certain days/nights/shifts.
And this new, young ER doctor starting out his career confessed coming up with the weekly schedule to satisfy as many of the doctors as possible was the MOST stressful aspect of his job. And this is an ER doctor we are talking about. Granted he worked in a hospital in a suburb so not as many shootings/etc. I thought that was hilarious to hear.
Ouch. I will say that the moment when I hit “send” on a new schedule is pretty stressful. Thankfully everybody is very understanding of mistakes, and nobody’s being forced to work a night shift in this case.
This is my research area, and this is exactly the problem. While technically these kinds of problems are np complete, they are usually easy in practice. Getting people's requirements is much, much, much harder.
Exactly. It took me a couple of weeks, working in my leisure time, to get the solver to work well. It took me a couple of years of feedback from my victims to give it a set of rules and weights that didn’t need constant tweaking.
Is a simple hill-climbing algorithm really effective for problems like this? I've found that you usually have to add some randomization plus simulated annealing to avoid issues caused by falling into local maximums or always iterating through entries in the same order.
Even local optima are often pretty good in practice.
Also, it's a very good practice to iterate through the neighbors in a random order (if you're doing next descent, for steepest descent obviously it doesn't matter).
The simplest version where you just choose the single change that produces the best score isn't quite good enough for my purposes. I modified that slightly to run that to completion, then switch to choosing the two changes that produce the best score until that can't improve anymore either, which works a lot better. Fortunately my membership list is small enough that this is still computationally feasible.
Real-world problems are so much messier and more frustrating than the clean abstractions many of us choose to work with. But this is really meaningful work that should have concrete social benefit. We should all hope to do good as you have.
Do you think more user and stakeholder engagement would have helped the public accept your research? I have no idea if BPS has the bandwidth for supporting this type of engagement, but I imagine a lot more fears would be assuaged if parents could have some role in "co-creation" of the algorithm, rather than just accepting results.
A lot of research from the fields of citizen science, public understanding of technology, and user experience show that participatory design techniques can enhance acceptance. E.g., https://www.bmj.com/content/319/7212/774.short
I am sure more engagement is always a good thing! One of the tricky tasks of BPS is to make sure that this engagement is "balanced", i.e., everyone's voice is equally heard. And you can imagine how hard that may be in practice. For example, Boston Public Schools had this survey asking people for their preferences on the start times. But most of the answers from these survey came from schools with less economically disadvantaged families. One way to do it is to call the families directly or have a lot of local meetings. But it is still a complicated problem.
Thanks for the response. I'm in the technology and policy program within IDSS and we have discussions about these "politics of technology" issues a lot. I'm sure you could become a case study in one of our classes if you wanted :)
Did you think the project was useful in the sense of getting the parents to think about possible changes and rejecting them? So they as citizens are clear that the current system is sub-optimal but they are OK with that
That's a great question, that I asked myself a lot.
I think this project was useful because it started this debate and brought more information to the table. For example, I found that a lot of people did not know how much the start times affected the district costs (tens of millions of dollars every year), or how the schools times seemed to be unequitable.
And this Boston Globe article allows even more people to read about this issue and make their own opinion. Ultimately, a lot of people in the US want later start times for high schoolers. This work can provide some information about the ways to implement such a change, as well as the consequences.
Did anyone suggest making the change, but announcing it today and having it come into effect a year (or even two) from now? That would give parents time to plan adjustments to their schedule, and would give a lot of people the chance to move schools if they really wanted to.
> District officials expected some pushback when they released the new school schedule on a Thursday night in December, with plans to implement in the fall of 2018.
The first round of applications for schools in Boston is January, so it's only a few weeks (over Christmas!) for families to figure out if they can make it work, if it's worth moving schools, which school they want to move to, etc. (Your chances of getting into your first choice school decline rapidly after the first round).
Or altering the ideal results to be less drastic at least initially to incrementally work towards the goal. Clearly the problem is the size of the change, not what the change is.
The engagement of the MIT team is interesting, but this reads to me like a classic failure of a "big-bang" rollout, rather than a gradual adoption. Was there any back and forth to create a slower-paced rollout plan?
We talked about it at some point, and it seems like a good idea. There are two main issues with it: first it can be hard to do a step by step change that is (a) fair (how do you choose the first schools to change?) and (b) works for every step (make sure that the transportation costs do not explode in between).
> it can be hard to do a step by step change that is (a) fair (...) and (b) works for every step
Yes. This is what separates the pros from the not-pros in major systems design and operation. The transition plan is the major key to plan, not the algorithm (which plays the role of Shiny Object in this case).
It is lots harder, lots, to get smoothly from A to B than to simply big-bang B. It takes enormous attention to detail and thinking through of every angle, and when done well is a thing of wonder.
I think in this case your team was not well served by how it was tasked.
Thanks again for responding, and for your team's efforts in the design.
One of the changes was moving start times of high schoolers later, which often came at the expense of younger students who would've had to start earlier. In my neighborhood, having to put 5 year olds on the bus at 6am or earlier was a big cause for the uproar from parents.
Can you talk about how you made this tradeoff? There's research that suggests high schoolers do better with later start times. But is there any research that backs up doing it at the expense of younger students? ie, is it possible that all students benefit from a later start time?
> Research findings suggest that changes occur in the “biological clock” during adolescence. As a result, teenagers have a natural tendency to fall asleep later and to wake up later. This is referred to as sleep phase delay.
> A correlation was found between subjects' melatonin secretion and their stage of development. The results indicated that melatonin onset occurs later in adolescents, making it difficult for them to go to sleep earlier at night. At the same time, the hormone “turns off” later in the morning, making it harder for them to wake up early (Carskadon et al., 1998, 1999).
It's like things are happening to the body during puberty.
I am no specialist of children health, but my understanding is the following: I guess starting too early is never good for any child, but there is something specific with teenagers that makes it a lot worse. There is much more medical literature documenting this fact than for the younger kids. So yes, in a way, given the constraints of costs, the algorithm had to choose the lesser of two evils.
Is it not possible to limit your algorithm's maximum time change for each school? Say, capping the change at 1 hour, and finding the best result of that set of times. Even if it is not as good as the 100% optimized solution, it seems like more palatable while also being an improvement.
Presumably, you are already doing that, by limiting the start/end times from 7am-10am, or something similar. In this case, instead of a global time restraint, there would be a local time change restraint +-1 hour for each school.
It seems the amount of change was a large problem. Was it attempted to minimize either number of student start times changed or total delta of minutes changed? Was it discussed how this could be done incrementally to slowly move towards more optimal start times?
Yes incremental changes looks like a good solution. I think it is hard to do it with the delta minutes changed, as you could have dramatic consequences on the costs (the ripples on the transportation systems can be quite surprising, in that changing the time of one school can have a big impact on the number of buses needed). As to the number of student start time, the main question is how do you choose the schools that change first? How do you make sure that your changes are not interrupted in the middle for political reasons, which would be unfair for the first schools and maybe the whole system.
Also, it is an algorithmic challenge to find a "path" of good solutions that go step by step to the good final start times.
I’m curious if an additional constraint like “total (absolute) time shift from this year to next (over all families in the district) <= K” or “max shift in first bell over all schools in district <= MS” was considered, along with the acknowledgement that this is e.g. an N year process to get from the current schedule S_0 to the optimal schedule S^*.
In my mind, this alleviates a lot of the “would have to change jobs this year” concerns, because the people making changes would ALREADY be making bigger changes (starting to take their kids to school), and the most extreme schedule changes would be deferred until the children on the existing schedule had aged out of that grade’s current schedule.
(Done enough optimization modeling to know this isn’t an easy path... but seems like it has some potential.)
Similarly, (maybe this was just the article dumbing down reality a bit, but it sounded like the transit expenditure vs parent happiness units were somewhat fungible. Any possibility of offsetting a bit of upfront expenditure with future savings of the more efficient (say from year N/2 onward) future state?
Thanks for working on the technical side of this social-equity project. Was the possibility of parent offering car pools or "walking groups" for kids included in the algorithm? I'm assuming that if volunteer parents are compensated to give carpool rides or lead walking buddies, then the negative impact would be lessened to parents who has to handle drastically different school start/end times. At the same time, it may be cheaper to compensate volunteer parents than for the district to procure or maintain more buses.
Another way to ask this question is: were there volunteer transit, ride share, or pre/post-school care incentives included in your algorithm, to balance the potential negative impacts to parents?
Yes and no. The algorithm runs with historical student data and therefore only considers students using the bus system in order to compute the effects of school start times on transportation.
On the other hand I think the district had planned to used the saved transportation money (~18 million/year) in order to help the most impacted parents (pre/post school care...)
I guess volunteer transit is always a positive thing for the parents and the districts. The algorithm was just trying to fulfil the objectives given by the district (costs, later students, fairness...). The compensations are what should happen after the algorithm: what do you do with the saved money, how do you help the most impacted parents? These negative effects were expected to disappear with the years as new parents start to choose the schools after the start times have changed (and therefore choose schools that can accomodate their schedules).
Not about your research, but any thoughts on SFUSD's assignment algorithm? I ask because it was also designed in consultation with MIT (in part) and also is an example of a very thoughtfully designed school district algorithm[1] that has been met with widespread displeasure in practice and is beginning to be dismantled.[2]
I think changing the start times is something the district had wanted to do for a long time. Therefore once the research and the algorithm suggested it was a good idea, they quickly jumped on board.
Now I am sure we have done some communications mistakes. For example the algorithm does not "decide" the start times. You just give it a (long) list of preferences and objectives (e.g., less HS students starting early, more fairness across the population...), and then the algorithm generates thousands of solution that try to satisfy these objectives. Solutions are carefully reviewed, and the objective updated if we think we can do better.
So in the end, this research is just a tool, it does not solve all the problems, and in the end people still need to find a way to agree. Hopefully they have a little bit more information to do so!
> start times is something the district had wanted to do for a long time
So, was the real motivation here one of efficiency and cost-savings then?
That is a clear thorn they might have wanted to solve.
Is the added sleep beneficial to students, all other things equal? I'm sure it is, but was the real motivation a cost savings and the sleep deprivation a nice cover for the real motivation?
No, as far as I know having later school start times for high schoolers was the main objective, and that was the strongest one in the solution that was chosen by Boston when they used our algorithm.
There has been a lot of research on high school students lack of sleep, and it seems that the school times do have a important impact on the students :
Did you read the article? It clearly stated that cost-savings was one of the outcomes to be balanced, but not the motivation. There were solutions that saved lots of money that were thrown out because they didn't balance the other outcomes enough.
BPS is pretty forward-thinking, and lots of research has shown that high school students are better off with later start times. Don't dismiss this as the actual motivation.
Did BPS consider applying the changes to only one or
a few high schools and their feeder schools? Perhaps
the realities of distribution of students among schools
in Boston not allow doing such a beta test?
That could be a way to do it, and Boston decided to do it all at once. I think their reasoning was the fairness of the process (it's hard/unfair to use beta testing on public schools). Also, it is such a hard change that maybe it is easier to do it all at once. You may also want to consider the political implications that are very important when making this kind of decisions.
Don't get discouraged, if you were feeling that way. It was important you substantiated the envelope of what kind of improvements are theoretically possible. And although the wealthier residents, who really ought to have the property to absorb inconvenience, may have been self-serving, they might reasonably have a point that sudden upheaval can cause problems for others that cannot attend meetings.
Maybe some will remember this if money for student transportation comes up to ballot.
My company had pitched the students of the A-Lab a couple years ago, but unfortunately we didn't get picked. BPS was one of the others pitching the students at the same event.
We submitted it to a journal that has a fast review process, and we are towards the end of it. We did not think soon enough to put it on arxiv, and we should be able to do it soon (it requires some editing to remove the journal-specific things).
The process went too fast to be entirely transparent. Boston tried to communicate as much as they could, and anything can be requested (Freedom of Information Law). We also have a paper that should not be too far from publication that details every detail of the thing.
I would argue that although it could have been more transparent, it is way more transparent than the status quo, where each individual school can try to pressure the district to get the time that they want, without considering the "big picture" and the impact on other schools/neighborhood/budget.
All decisions were made after a lot of group conversations with parents/staff/principals... But I am sure some things could have been better, in the end we were just the grad students trying to help :)
This can happen and did happen before. Maybe being suspicious of an algorithm is not a bad things, we should have extremely high standards for them ! My hope is that some people will just have a better view of how complicated this problem is, and be able to construct their own opinion from there. And this would already be something I would be proud of!
Meanwhile, rich parents work jobs with flexible hours and poor parents are hourly with potential catastrophic penalties for missing work. Theoretically, you're correct (from a parent preference perspective, at least). In practice, even the rich parents would balk at the dollar amount required to make it worth it.
That also completely ignores the fact that what you propose would deepen the existing structural inequalities for students, and presupposes that schools are and should remain stratified by income. I vehemently disagree with this unstated premise.
One of the reasons that this would be a terrible idea is that the person who stands to make money by selling their good slot for a terrible one is not necessarily the person who has to wake up at 5:00 AM to bus across town.
Adding money into the mix, when there's a large difference of means between families, adds a level of incentive that shouldn't be there -- one that has nothing to do with the district's goal of providing an equal education to all its students.
When the district looks at the results at the end of the year, and all the poor kids have poor grades/attendance because they were all in the worst time slots, they can't say "well, at least the parents are slightly richer because of this, from the secondary trading market." Rather, based on the objectives they are concerned with, they may have just made inequality even worse.
Ah yes, I'm sure the whole thing would have worked politically if only they'd also tried to set up a system in which rich parents paid poor parents in order to get better start times. Conveniently, this would also mean that the students would be income-segregated so rich students would not have to be mixed in with poor students.
</snark>
... I also took economics in college. To me it seems obvious that calling a fanciful solution like this "obvious" except for "human pride" is woefully inaccurate.
That sort of arrangement incentives people in many unexpected ways, like people optimizing for the chance to get allocated a slot that can be sold at maximum price, rather than what is the 'best' for them overall. It then turns into a game theory exercise of course.
This is a huge overreaction - a sort of caricature of the slippery slope argument, attacking points that the parent did not make or even remotely imply, that was presumably intended to signal virtue, but does more to discredit is maker.
Distasteful as the parent's comment might be, they are talking about Pareto optimality - finding a situation where both parties would agree it's an improvement.
>From an economics standpoint the ideal labor force is slavery.
How so?
A happily paid worker with their own free time isn't likely to revolt and kill you. Even an unhappy worker who hates you will just quit. Is your calculations forgetting to count such costs?
And a machine that replaces a person is far far more efficient. So machines backed by human operators/fixers/etc. end up outperforming humans, and free humans will have far better education possibilities.
>From an economics standpoint there is no point to the elderly they should be killed off once they are no longer viable laborers.
You are clearly zeroing out the cost of having all non-elderly realize this will be their fate as well. That seem an arbitrary cost to zero out and makes the results of your calculations unusable.
>Children are only useful if they show potential slave labor ability
A free child that is educated has a much higher return on investment on average.
>otherwise should be killed in adolescence if there are obvious shortcomings
Once again completely ignoring the costs of taking such a course of action.
I'm not really having any confidence in your models.
> A happily paid worker with their own free time isn't likely to revolt and kill you
Historically, slave revolts are extremely rare. Most slave populations are deliberately beaten into a state of learned helplessness so that they don't even believe that revolt will get anywhere. Or they are given just enough such that there is something to lose (besides their lives) in revolt, and the natural human aversion to loss kicks in. Many slave economies in human history have relied on this - see ancient Rome as an example.
> A free child that is educated has a much higher return on investment on average.
But that ROI takes 20 years to be realized. How many people, really, are going to forgo ROI today for bigger ROI two decades from now? There is a reason why on a normal yield curve, loans with long term maturity have higher rates of interest - psychologically, we are biased towards present income vs future income.
> How many people, really, are going to forgo ROI today for bigger ROI two decades from now?
That's (part of) what governments are good for! They are great for making decisions in situations where larger investments over longer time periods pay good dividends.
At one point he discusses the wages in North America, and how high wages seem to correlate to rapid economic and demographic growth. That does seem to mean he doesn't quite agree with you on what is more efficient.
In other places, Adam Smith seems to think slavery is rather inefficient. One argument he presents is: who do you trust more with the upkeep of a person?
* a master who may have many priorities;
* or the person themselves, where their upkeep is by necessity their own highest priority?
While I sympathize with the aim of your comment, it seems factually wrong to me.
In the past 100 years, it seems that individual freedom and economic prosperity of the society as a whole correlate. It is hard to tell what is the cause and what the effect. But it pretty obvious that - on average - less free societies are not very successful economically.
With respect to the elderly, if I would be killed on retiring, my incentive to work hard and generate wealth until then would probably not be very high. I think that - not only, but also from an economical point of view - it is much better to let people pay into a pension scheme instead.
And all economic experts agree that a robust legal system is much better for economy than murder, war, and violence.
Economics is the study of choice. I think you're confusing economics with some straw man caricature of libertarian capitalism.
Utility maximization is a core concept in economics. The definition of what utility is can vary. Utility isn't equivalent to money. Anything quantifiable - and you can quantify qualitative things if you give them some dimensions - might be input to your utility function. What economics does is gives you tools to reason about the systemic effects of decisions that balance competing utilities, and how to apportion scarce resources.
If you're trying to contend the study of economics is not highly politicized, then in my opinion you have your head in the sand. Or that economics is ad reductio absurdum to what you described and nothing more, then you must be oblivious to the world in general.
What you are really doing is just parroting what laissez-faire trickle down econ people... oh wait, by your definition they wouldn't be commenting on something so broad, macro, sociological, anyway they parrot that junk as if the redirection of wealth to the very few is some mathematical law, as if the pirate distribution game theory actually applies to a massively complex chaotic system like the real world.
This article doesn't mention it, but the biggest problem with Boston Public Schools transportation is that students don't actually attend the school in their neighborhood. The school district assigns you a school that may not be located particularly near you, in order to increase school diversity. So a lot of this transportation time and expense is taken up to shuffle students around to far-away schools.
Denver Public Schools have a 'choice' model that works alright[0]. You can opt, for free and without restrictions, to go to another school other than the neighborhood one (you are automatically put in the 'closest' as a placeholder). Generally, there is no busing with DPS, as school kids ride free on public transit. The Denver School of Science and Technology[1] is a 'charter/public' portion of DPS that regularly pull from across the metro, mostly via public transit. There are many other issues with DPS, but the 'choice' model they use, along with the assistance of public transit, generally works pretty well.
New York City also uses citywide school choice for high school. School choice really doesn't scale up. You end up getting a yellow-pages sized book of schools (and each page is a single school!), you trek across the city to view schools, there are entire forums and such dedicated to navigating the process. It's like the college application process in miniature.
We also generally do not use busing, but New York City is so large and congested that you start asking a lot of kids. I went to a school that was 90 minutes away by public transit on a good day, so I'd wake up at 5am to make an 8am start time.
Oh man! A friend in HS had swim practice before school and woke up at ~4:30 for it 3 days a week. On each of those days, she was just a zombo. I can't even imagine doing that every year. Do you think it was worth it for you?
The other choices, unfortunately, weren't much better. New York calibrates its schools to match up with parent commutes so that kids leave at the same time or earlier, and these parents in my neighborhood often had the same 90+ minute commute. As an example, there was a much closer in school that was 45 minutes away by bus, but they started at 7am, so it would've required an earlier wakeup time.
There was a school within walking distance, but the graduation rates weren't too hot. The nice thing about school choice, in theory, is that it doesn't cause neighborhoods to gentrify or disintegrate due to the quality of the local school. But if you scale it up to a million kids you basically make it so that only people with time and money can really game the system. I wouldn't consider myself particularly blessed, but I definitely was not the worst off person.
Here in Seattle we used to do the same as Boston, but a Supreme Court case where a white student claimed discrimination due to the process used to do school assignment ended it. Now we have neighborhood schools again. Although even with the neighborhood schools there is a lot of angst since often more affluent neighborhoods will push their way into a better performing school, even to the detriment of neighbors who are closer to said school.
Coincidentally, Seattle, two years ago, also implemented later HS start times and earlier elementary start times. Our elementary child went from a 9:30am to 7:50am start time. It sucks and there was some outcry, but like so much in Seattle -- there was mostly just a collective shrug.
I don't know if I'd really characterize that as a problem though. I did a lot of research for this a while back and a lot of research has been done that shows demonstratively, one of the best way to dramatically improve educational outcomes is integration (both economically and racially, though the two are many times closely linked). And because of many factors that almost always involves bussing students to schools in different neighborhoods than the ones they live in. Despite the narrative that bussing in the 70's and 80's failed the honest truth is that all the studies showed that it worked until the more affluent white parents fled to the suburbs (for good and bad reasons). Integration of schools is inconvenient and it takes time for everyone involved to adjust but the research is pretty clear that it is overall a good thing for pretty much all parties involved.
the more affluent white parents fled to the suburbs
it is overall a good thing for pretty much all parties
These two statements conflict with each other. If it was overall a good thing for all parties, you would not have people fleeing to good school districts.
Yes, Boston Public Schools has the second highest transportation budget per student in the country as a result. If I remember correctly it is ~25$/student/day. But one could argue this is the price to pay for increased diversity.
Is that $25 per student overall or per student who uses school provided transportation services? If it's the former that is about a quarter billion in transportation costs annually.
Do you know off-hand how this compares to the rest of the system? I mean non-student public transport... maybe just busses? Per-rider and overall, ideally.
These are the figures from 2015 [1] Subway/bus fares are $2.75 each way (obviously lower with a monthly pass) which covers about 30% of the total so the cost per rider is ~$8 per trip. The overall figures include commuter rail but I'm just assuming that doesn't change the percentage covered by fares all that much. The MBTA carries about 1 million riders per weekday.
This, essentially, looked like a grander scale version of almost every web design project ever. What is different is that, over time, web teams have learned to use "mockups" or "wireframes", to put in front of the end recipient some simulation or approximation of what the final result will look like. Why? Because otherwise you get this sequence:
1) end recipient tells you what they want
2) you build that, perhaps exactly
3) they don't like it
Thinking in abstractions, and imagining what that would really look like once implemented, is more or less a job description for programmers. It isn't, for most other jobs, and thus most other people. If you don't do frequent reality checks during the process, you will be quite likely to get this result at the end.
Of course, the students doing the programming in this case, were unlikely to have had multiple software projects' experience, to know this. But someone there should have, and should have insisted on putting sample results in front of parents to see what their theoretical response would be.
I find that management, and probably also politicians, don't like doing that because they can't control how the conversation goes, and it could bring them bad news. But putting that off until the end, brings worse news.
Isn't that more or less what they did? Having a couple of grad students compute exactly what shifting high school later without costing a lot of extra money sounds sort of like the definition of exploring what the change would look like.
ADDED: To be fair, they did try to take the results and try to run with them. It seems as if there would have been less of an issue had they taken the results and did more outreach to find if this was a change parents were actually OK with.
Yeap, the article has a very heavy handed one side to it.
I mean, it sounds like "Silly users don't like my smart product that no one uses" without considering, maybe, just maybe, the product wasn't good?
The author did touched on it a little by mentioning the difficulty for working parents to move their work schedule, but it seemed glossed over. I'd think changes that interfere with the the household bread earners ability to earn would be a serious consideration and a deal breaker for any proposed change.
Not to say politics aren't incredibly inefficient and way more emotive than logic prone.
Children who live too far from the school to walk catch a bus, an ordinary scheduled bus on a standard route, or the train# or the metro. They get special bus passes (some get them free, I think most now).
What I'm saying I think is why on earth is the Boston school system in the public transport business to start with?
(#) Doing your homework on the train on the way to school is one ritual I was amused recently to see is still going strong. I did this and my dad did this (pre WW2).
In order for a public transit system to be adequate to transport children to school, it has to be adequate, full stop.
In the US, a separate public transit system specifically for schoolchildren exists, usually because there is no effective public transit system for everybody. A child cannot catch a city bus from their home to their school when no city bus ever comes within 3 miles of their home.
Europeans often don't understand, because American urban planning makes it seem like one would have to be a madman in order to do things that way. It's not madness. That's just the way the economic incentives aligned in the past, and no one has cared to re-break the bones in order to set them properly, so everyone is crippled from birth by the history of the infrastructure.
Your point is valid but the person you replied to was specifically asking about Boston. I admittedly don’t know a lot about the public transit infrastructure in Boston but I have a hard time believing it’s less extensive than e.g. Seattle, where public high school students use the standard bus system, with the school district paying for special limited routes where the normal bus routes lack coverage.
Fair enough. I have never lived in Boston myself, so I can't speak to its specific issues.
Some school systems in the US are subject to judicial desegregation orders, so that the racially-divided neighborhoods do not result in segregated schools. These may require transit routes that are not necessarily coincident with the daily work commute. Some systems also have magnet or specialty schools. Some have even greater ability to choose (or win via lottery) a school other than the nearest to one's home. As a result, the commute for one child could be longer than that for their parents, and using the regular public transit could result in hours on trains and buses. In such cases, a school bus would essentially be an express, with no stops between school and the residential neighborhood.
And, of course, once you run one school bus, that means you now have a school bus infrastructure, which makes it easier to run more buses.
The public transit system is usually designed around moving people between workplaces, mainly in the city core, and homes, which are more evenly distributed. Schools are also more evenly distributed, as villages, towns, smaller cities, and unincorporated townships may have had one or more of their own before being annexed into the metropolitan city. As a result, the nearest transit station to both home and school may be the same station. Or they could be on lines that run nearly parallel before meeting in the downtown hub.
One possible solution to this would be to build a large, unified school near the city's central transit hub. And that would likely be selected by an algorithm as most efficient. But politically, people prefer their schools to be near their homes. And separated by grade levels. With athletic fields and playgrounds. And lots of separate and distinct administrative regions. Again, the technically sensible solution is untenable, because people just want what they like (and what makes them money).
Boston has a pretty good public transit system--at least until it snows when things tend to go to hell. It has its share of problems in general but it's not bad overall.
Public transportation systems in the US are mostly focused on getting people to and from employment centers and therefore don't provide the transportation coverage needed to get students to schools.
Also in the US once you are outside a medium to large city public transportation may not be available at all. Or is limited to systems which run only during commuting hours and are designed to transport people from a few centralized locations into the closest city.
The Boston public schools don't operate on the principle that you go to your local school. Kids get sent to quasi-random schools so that you don't have the good school in the rich neighborhood and the bad school in the poor neighborhood. Instead you get a whole bunch of medium quality schools. Also, this means that children are very likely to not live near their school at all.
do you at least get to stick with one school until you've completed the respective grades in that school? also - do siblings end up in the same schools? or are they also randomly assigned?
For high school at least it's like applying to collage, you take a standardized test and send out applications which some schools (Boston Latin) being more desirable then others.
I'm pretty sure that when i was at school in the UK, there were specific school bus routes. They weren't operated by the schools or the local education authority, so i assume they were special services run by the local bus operators, using stock that would be used on other routes at other times.
I believe each route served multiple schools - kids from my school would share the bus with kids from other schools. Since we were the grammar school, full of nerds in garish purple blazers, and they were, well, not, this was sometimes a stressful experience.
It may be relevant that this was in Colchester, a town with numerous schools, some quite close to each other, whose catchment areas take in lots of small villages, where there isn't necessarily a good normal bus service.
Mind you, i got the (ordinary!) train, so i'm not sure of the detail.
I came here to say almost the exact same thing. From age ~10 onwards, I would walk to school with my younger sibling in order to get to school for 0850. When I moved schools in year 7, I left slightly earlier to accommodate the change in distance as the school started at the same time, and my sibling walked alone (they were then ~10). My parents went to work whenever they needed to.
I could alternatively have taken a (public transport) bus, but the fare would have come out of my pocket (i.e. out of the small monthly allowance I was given to spend on whatever I chose).
I was asked to choose someone to receive a small prize/incentive, out of a list of fundraisers. I'd never done anything like that before, so I thought "I'm going to do this right!". So I went to random.org to get a real random number to select from the list.
Alas, my first selection was rejected by my supervisor because they were related to a co-worker. Ok, try again.
The next selection was rejected because she knew they weren't able to utilize the prize.
The next selection happened to be someone who had raised the least of anyone on the list. That wouldn't do.
I finally said "ok, you choose someone that's deserving and hasn't won recently".
Computers have a way of exposing you when you think you know what algorithm you want, but you don't.
This is nothing to do with the algorithm, and all about the downsides of centralized administration.
If anything, it's just a case of political leaders failing to communicate and take action to garner support for doing something with abstract/non-existent benefits and concrete, likely, and easy to measure risk. They did use the arrogant and paternal approach of big tech company, though.
Based on the figures shared in the article, it smells like the district had an objective of saving some cash, while getting some attention for saving high schoolers from sleep deprivation. To meet the goal you have a few strawman arguments that would waste time or money, then settle on a solution that met the cost target, using difficult to argue with rationale like racial equity as an justification.
It would have made more sense to slowly introduce changes, and perhaps measure the stated outcome before disrupting everything. Given the really high cost of daycare in Boston, it seems unconscionable to force parents into situations where you need a much larger block of afterschool care.
How little the health of students counts compared to the whims of their parents.
When did it became a right for parents to put their kids on a bus? They wait on the sidewalk, the big yellow truck comes by, they get on it. It's not that hard. I managed to do it without parental intervention every day from 2nd grade on (legally required before that where I grew up), because my parents' work schedules conflicted with the bus schedule.
And I'm not some old-timer complaining about kids these days, I'm < 30 and I still remember being completely unable to fall asleep before 1am as a teenager and being expected to wake up at 6:30 and go work for 14 hours a day. On behalf of their kids, fuck all of these parents.
> Sorting through 1 novemtrigintillion options — that’s 1 followed by 120 zeroes —
"Trying 10^120 options..."
We can stop padding our writing to make page length and column inches now. Digital media has a page area that is exactly as large as you need it to be.
If you're a journalist, even in long form stories, your working vocabulary is essentially the 3000 most-used words, and anything outside that should be easily understood from context alone. If you have to follow a word with its definition, you need to write something else. I don't know who edited this, but their iron palm of author-correction should have started the smack after "novem" and finished it before reading to the end of the "illion". A large number doesn't need a large word to describe it.
The essence of the problem is this. Balance the following factors:
(1) Supervise the children of working adults, without interfering with their work day.
(2) Transport those children between their homes and their schools, at minimum cost.
(3) Teach the kids how to be working adults after they graduate. (3a) Prepare a fraction of the kids for further education, to associates or bachelors.
In light of research into school start times as a factor of academic performance, they asked some MIT grad students to improve on goal (3) by giving high-school students later start times, with minimal impact to (2) from the new bus schedules. The algorithm devised to do so failed to sufficiently account for goal (1)--with all of its obfuscated hidden requirements--which made the whole plan fail.
They simply forgot, or never knew, that the primary purpose of public education in the US is to allow both parents to work outside the home, and to subsidize companies that have extended business hours with state-provided supervision for the children of their employees--a benefit usually paid for by the employees themselves via the property taxes on their homes.
The technical solution is easy. The political solution, which includes an admission that this is one of the goals baked into the public schools system, is nigh impossible.
I think "novemtrigintillion" was just a harmless joke about this very large number of possibilities, nothing too serious about it, I also had to Google the word I doubt anyone knew it.
> (1) Supervise the children of working adults, without interfering with their work day.
That is indeed an important objective. I want to underline that it is impossible to achieve it in practice. Each parent has a different life with their own constraint, but each school can only have one time. The complexity of school assignment and start time choice is to find a way to do as well as we can. Unfortunately to the best of my knowledge, there is no perfect solution.
That's not strictly true. The schools each have one time when mandatory attendance starts, and one time when mandatory attendance ends, but they all open their doors earlier, and close down later, when the student presence is still permitted. The bus schedules revolve around that 90% of kids that show up just for class and then leave for home afterward, but if there is a local need, it is usually technically possible to fill it. It just takes money. The problem is that the need and the money only rarely intersect.
Some school cafeterias even serve breakfast--mostly for kids with subsidized meals, who may not be able to eat regularly otherwise. That is state or federal money available to open earlier. Youth diversion program money can keep a school open later. It is entirely possible for a kid to be at school continuously for 12 hour intervals, every weekday, with no additional burden on the local taxes. Meanwhile, in the suburban schools, the clubs and extracurricular activities are powered by fees paid by the parents directly, and some kids get picked up and dropped off by private cars, as part of their parents' commutes, never even seeing the inside of a school bus.
You're right that public schools in Boston and elsewhere benefit from a rich ecosystem of after-school programs that allow students to continue learning outside the classroom (and yes, give parents more time to work). Adjustments to school schedules also impact these after-school programs, especially ones that enroll students from several schools with different schedules.
Exactly. I was working on some machine learning code when I realized that there were implicit moral choices in all machine learning results. Every dataset has implicitly made choices about data to include or not include. Every "weighing" algorithm, even if emergent, is choosing "values" to weigh or not weigh. Now I see data in a whole new way - and find that I can easily dismiss most conclusions, since they represent value systems which are very distant from mine. (Most often, they represent a value system based on profit - effectively keeping others down rather than lifting others up. How many people even noticed that costs were a major consideration of the article? We are so accustomed to our value system of profit over people that we fail to look at the system as a whole - and consider the incredible wealth of our society and wider impacts of actions.)
Most people don't get it as they don't understand reason behind their "reasoning" - usually driven by feelings, instincts (biology) and default modes they learned from their parents, friends or at school. So implicit choices made for them already ("common sense") cover the reasons about what really drives them and these are completely out of their radar, but we have to cope with them when we want to get anywhere near automated intelligence and morality.
This is very easy to see when you consider morality to be (perhaps by definition) any process which compares competing possibilities and labels some superior to others. There is an implied morality acting under any goal, because one always considers the states that achieve the goal superior to ones that do not.
With humans, we can have extremely complex goals, which leave room for our morality to shift around as the context changes in subtle ways (to make exceptions to rules, for instance.) Machines tend to have very simple goals by comparison. It makes it difficult to encode any potential exceptions to the rules.
It's not mentioned in the article, but Julia was used to solve the huge optimization problem for this project—specifically, the JuMP optimization package [1]. This SIAM article has more technical detail: https://sinews.siam.org/Details-Page/a-school-bus-trip-to-th.... This seems to be a preprint of the paper that may describe the algorithm used in Boston, although it seems to talk more about the well-known New York taxi data set: http://web.mit.edu/~jaillet/www/general/travel-time-18.pdf.
One thing many programmers fail to realize is that almost all problems in the world are ultimately political, not technical. Data analysis is almost always a tool, not an answer.
The hardest job in the world is getting a large group of people to all do something. The way you message something can be as important as what you say. If I was working on a major schedule change project like this, I wouldn't even tell people about the algorithm. That's just a tool to help you arrive at a schedule. People don't want to feel like their life is controlled by an unfeeling machine even if the result is better for them.
I'm not saying this particular situation was winnable. But I've seen this kind of thing happen over and over in the tech world.
I agree that the role of politics is important, and maybe it is for the best if it can help create a conversation.
One of the main objective of this particular project was to make a complex problem more accessible to a general audience, so that such a debate could take place. For example, before this "algorithm", it was harder to understand the impact of your school time on the other schools and the budget of the district. And this budget is also what pays for teachers, nurses and buildings.
My hope is to convey that yes, algorithms can be scary when they decide for you, but sometimes they are just tools that bring information to the debate and help the minorities be heard. And ultimately, humans make the important decisions.
Thanks for your insight. I think the work itself was really interesting and super useful. My point is just that often great work is a tiny component of successfully convincing a semi-rational and biased populace of agreeing to something. It's just plain hard.
I hope you'll keep on doing great stuff like this even if sometimes the surrounding political situation outside of your control is messy and sometimes feels discouraging.
Certainly there are an awful lot of startups that fail over issues like this. They either invest massive effort into seeking technical fixes that don't exist for human problems, or they find genuinely superior technical answers that are impossible to achieve - because they upset key parties, or because the cost of change is too high, or because they just sound too damn silly to sell people on.
But in this case, I'm not sure there was anything to be done. (As you acknowledge - I'm not disagreeing, just looking at the point.) Two grad students whose background was overwhelmingly technical were brought in to provide a technical solution. They had enormous amounts of logistical information, but it's not clear they had social/political information about their task, and certainly they had no control over messaging and notification.
The outrage appears to have been less about the use of an algorithm than about massive disruptions to schedules. It's a predictable outcome - school start times are largely set around common parent work schedules - but it's also an irreducible one. The desired outcome was the source of the anger, however much the computer derivation added insult to injury. Doctors have been calling for these changes for years and been utterly ignored. The school district I grew up in tried something a lot like this a decade ago, with no fancy algorithm, and was slapped down with similar fury.
At a certain point I think the error was not the superintendent seeking a technical solution to changing start times, but seeking a solution at all on a topic where parents didn't want change.
(If I were to really get inflammatory, I might mention that school is a classic principal-agent problem, and the proposal failed because it tried to trade the agents' convenience for the principals' health. But that's another, uglier debate.)
>(If I were to really get inflammatory, I might mention that school is a classic principal-agent problem, and the proposal failed because it tried to trade the agents' convenience for the principals' health. But that's another, uglier debate.)
Or, at least (to phrase it in a slightly less negative way) to trade relatively incremental, diffuse, and somewhat abstract benefits for the students with specific and sometimes disruptive change for parents.
Furthermore, just about any change to an accepted status quo--even if it's arguably a mild net positive--is going to have a lot of vocal critics. Those who like the change will mostly not feel very strongly. Those who don't like it or have just learned to live with the status quo will often hate it.
> At a certain point I think the error was not the superintendent seeking a technical solution to changing start times, but seeking a solution at all on a topic where parents didn't want change.
Agreed. It's a super hard problem to solve. But this situation as a whole I think is still a good case study that developers should consider. You can do the best job possible and still get "blame" because of the political situation beyond your control. You can't fight politics and emotion with data alone.
That's why the smartest people should shut up, create a great technology that solves issues people can't agree on for centuries, then entice young smart generation to adopt it and then overrun the issue in the real world by being the only solution. Once politics and interests get involved, it's over.
I'm pretty sympathetic to this tactic - it's amazing how many 'intractable' political and social debates have been solved by rendering the whole problem moot for a group that's not already entrenched. Science rather famously "advances one funereal at a time", and most debates over resources don't get settled so much as growth and technology render them uninteresting. (Recall that brightly colored fabrics were once a source of enormous wealth and even legislation.)
But I'll be damned if I can work out how to apply it to childhood education. It's almost definitionally a system run by the old guard, on 'behalf' of people too young to change anything. A lot of my friends and I got to our late teens and twenties, swore up and down we were going to fix some of the obvious idiocies and cruelties and wastes of education, and just utterly failed. There's no population to offer a competing system to, and very little road into the existing for any member of the 'young smart generation' which lacks either children (to be a stakeholder) or multiple degrees and time in the trenches (to be a participant).
I'm not surprised it failed, with such drastic changes. Why couldn't they add a constraint of no more than 15 minutes change and reapply every year for 5-10 years until they reach a point they are happy with?
Because done that way (in addition to what sibling comments suggest), halfway through, there'll be a point when everyone has to get to school at the same time and the bus fleet will have to triple, at huge cost. Or you just end up doing a large disruptive jump anyway to avoid that timing issue.
Of course they could, but as the article mentioned toward the end, the school district no longer has the political capital to pursue rearranging school schedules. The superintendent lost his job over this. I imagine votes were angry with the school board.
Repeated change has its own substantial costs for a system as large as a school district - even without the outrage over the drastic change, reworking schedules every year would be enormously unpopular.
(Think of everything from changing heating and AC schedules to reworking start times for bus drivers to telling parents they need to drop their kids off at a different time each year. There's a reason most school administrators I know get an absolutely hunted look when you mention changing anything.)
What are the downsides of having middle and high schools start at 9am with elementary schools starting at 7:30am? Basically a reversal of what is in place today.
Just go through with one fell swoop and swap every single starting time.
It may seem feasible on the long run. Indeed, it can be easier for the parents of the new children to adapt their schedule.
But it can be very hard for the parents that previously chose the school (partly) because of its start time, and adapted their own life & work schedule around this time.
A "two-tier" system with 7:30 and 9:00 starts is interesting - one thing to consider, though, is that many, many families in Boston would like for school to start between 8:00 and 8:30. So a solution with zero students starting in this window might be unpopular.
Also, it'd have a huge side effect: middle/high schools kids can go to school on their own. Many kids at elementary schools cannot. With this, you force parents to drop their kids off much earlier, which would fuck with their work schedules.
I'm from indiana. 7:30 is a common starting time.
I'll add that most parents in the areas I lived didnt take their kids to school. A lot kids take the bus, either from home or from a daycare. Additionally, many schppls offer before and after school care, usually starting at 7 or 7:30 for the parents that will pay for it.
Indiana has the most complicated time zone history of any state in the US because of a long-running battle between those who want to be on Eastern time for synchronization with the East Coast and those who want to be on Central time because that’s where the state’s longitude puts it.
Yup. As far as I've witnessed, syncing with either Ohio/Eastern time or Illinois (Central time) is the main bits of the matter. Northwest Indiana is pretty in line with Chicago: Terre Haute (southwest-ish) more on Illinoise. Lafayette (central west/Purdue area) is in between. Central Indiana doesn't matter so much due to Indianapolis. Southeast is more in line with Louisville and the entire eastern edge, once you are outside of the easily Indianapolis area, is more reliant on larger Ohio cities. (Cincinatti, Cleveland, and to a smaller extent, Toledo). Ft Wayne I'm not as sure about, though I think they'd rather stay in line with Ohio's time.
I*m not entirely sure why it is a big deal: Either way, it is only an hour at this point. To be fair, it was more difficult before indiana adopted Daylight Savings time.
Same in Italy, all schools - independently from grade/type - start in that timeframe, but there are some advantages from what seemingly happens in Boston:
1) Elementary and middle schools are (more or less) neighbourhood based, it is an exception that a pupil comes from far away.
2) The "problem" comes starting from high school, in Italy we have two "generic" high school types, the "Classico" and the "Scientifico" (I think there is no need for a translation) of which there are many + a number of "focused" or "professional" schools like Geometri, Ragionieri, Linguistico, Artistico, Alberghiero (Surveying, Accounting, Linguistic, Artistic, Hotelier) and usually it is the kids going to these latter types that have the longer commuting time/earlier waking needs, particularly those coming from villages nearby the city.
In my times, I was lucky enough to be able to walk (accompanied or later alone) to elementary school as it was less that one block away, as well walk (some fifteen minutes walk away) to middle school and go by bike or moped or bus to high school (a fifteen to thirty minutes trip at the most).
But I had friends that went to that or the other "professional" institute that had 1 h - 1 h 30 trips every morning, so they usually set the alarm clock to 6-6:15 in the morning.
Now I will sound as an old grumpy bastard, but all in all none of us suffered from the scary consequences outlined in the article:
"Sleep-deprived teens are at increased risk for poor academic performance, binge drinking, and suicide."
Maybe, just maybe, it is that risk that is a tad bit exaggerated?
First class at my high-school (Fairfax County, mid-90s) was 7:20am. Some still start that early, though the county has moved some to later starts. And some primary schools were moved into the early slot. There are only so many buses/drivers to transport kids.
Also, importantly, you'd then have younger kids getting out of school much earlier, which would require parents to pay for after-school care. And parents with younger kids are often younger and therefore generally making less money...
Allegedly the problem was that parents had to quit their jobs to adjust to the schedule. To me this seems easily solved by having little Jimmy walk or take the bus to school. No need for parental involvement in those cases as 12 year olds are capable of getting off and on a bus without supervision
I used to work in Boston Public Schools, so I have some context that may or may not address your point. One issue is that for the most part, Boston does not do "neighborhood schools" anymore. So a kid could easily have a 1.5 hour* walk to get to school. However, your point about busses largely stands because, at least for high school kids, the district mostly partners with public transit to bus the kids around.
* Downtown Boston is tiny. But the actual official city is pretty big, and dips much further south than many people realize.
It is more because older children can watch the younger ones after school. A 15-year old can watch the 8 year old for an hour or so after school. the 12 year old might be able to, but can definitely stay home alone. An 8 year old needs care. The later start minimises this with a "traditional" work schedule.
Im sure google has precise answers and it has changed since it became a standard arguement 30+ years ago, but I do imagine plenty. 3 years between kids and having 3 kids gets you pretty close (the 8 year old would have a 14 year old sibling). I personally have siblings 6 and 11 years younger than me. The bigger issue with me answering is that not only am I 40, but I'm childless and living in a different country than I grew up in. I*m not sure what the standard family actually is right not in different parts of the country.
Parents are a really tough constituency, especially the wealthier, former yuppie, homeowner cross-section. If you look at any major initiative to transform urban living, from housing to transportation to schooling you will find this demographic resisting with force.
I'm surprised that the school district did not foresee this resistance.
I think something important is being glossed over by this article: "...distributing the best school start times more evenly..." Can someone explain to me what makes a start time "best"?
There was a section describing parents whose elementary school kids would have to start earlier, and they didn't like that because the kids would have to go to bed earlier and it would be tough to rearrange their work schedules to accommodate. But won't this also be true when their kids are in high-school? If they can't cope now, how will they cope then?
I always though it was obvious that elementary school and high school start times should just be swapped because in my experience elementary aged kids always wake up far earlier on their own.
"Best" is a combination of...
- minimizing cost to the district (number of buses, etc)
- high school starting later in the morning (adolescents need more sleep)
- ensuring adequate child-care options are available for little kids. Early start means more PM care, late start means both AM and PM care, etc.
Then you get compounding factors...
- high school athletics. Parents don't want high school to start too late, because then there is no daylight for football practice (yes, this is an actual argument I've heard... parents would prefer their kids play sports than maximize the kids education).
- race and income issues - either correcting existing disparities or avoiding creating them.
> "Best" is a combination of... - minimizing cost to the district (number of buses, etc)
No. The paragraph I quoted implies that the parents have some notion of "best" start times that has nothing to do with the district's perspectives. But the article makes no attempt to describe what that is! (Other than what I already mentioned and questioned.)
Child-care, and not having your kids practice football in the dark is the kind of thing that makes sense.
But what does school start time have to do with race and income?
The paragraph I quoted implies that the parents have some notion of "best" start times that has nothing to do with the district's perspectives. But the article makes no attempt to describe what that is!
My best guess... people are generally change-averse and parents are generally selfish a-holes. Lacking irrefutable evidence that the new time is better FOR THEIR CHILD, parents will violently protect the status quo (even if it's obvious to outside observers that the district as a whole would be better off with the change).
> If they can't cope now, how will they cope then?
Elementary school students (especially in the early grades; it's less of a problem for the 10-11 year olds) typically need someone to help them get out the door in the morning and someone to meet them when they come after school. This introduces a coupling between school start/stop times and parent work schedules.
Your typical high schooler can get themselves to and from school on their own and doesn't need to be watched over when they come home, so their school schedule doesn't really have to match up to the parents' work schedule.
The article was making it sound like later is better. Maybe because that way you can drop the kids off on the way to work and then have more work time in the afternoon while they're still in school? But I suspect it's more likely that whatever time it is works because you set up your work shift schedule around it or whatnot, but you don't have control over _changing_ that work shift schedule, so changing the school start times becomes a problem.
This is obviously not an issue for people with flexible enough working hours.
I've experienced this pushback against the algorithm firsthand. I work in NLP, and I was asked to write a script to select data for annotation based on a number of criteria, including some automated balancing of a few factors. It turns out, this was very unsatisfying to some. It led to questions like "why is this document being selected, but not this one?" without an answer that was really satisfying. I ended up changing it to give total control to the person running the script, and removed the self-balancing feature. Since then, I pushback against any sort of black-box auto-selection sort of ideas, because I know that, while they might result in faster and ultimately equivalent results, people want control.
People don't want control so much as those answers you yourself didn't get.
Nothing is more infuriating than not being told why you're being punished, why someone got accepted into a school when you didn't, or why a judge won't consider your appeal.
We can even handle injust answers-- "his parents are alumni that donate a lot of money." That sucks, but it's the answer.
But we can't handle not knowing. Denying people the ability to reason is one of the cruelest things you can do to them. The Amish figured this out centuries ago.
The constraints of the problem seem to have been poorly specified:
"...the algorithm would eventually propose — shifting school start times at some elementary schools by as much as two hours. Even more.
Hundreds of families were facing a 9:30 to 7:15 a.m. shift. And for many, that was intolerable. They’d have to make major changes to work schedules or even quit their jobs. And because their kids would have to go to bed so early, they’d miss out on valuable family time in the evening."
A major portion of the problem is that schools aren't just about educating children; they're also about warehousing them as public daycare for workers. The hours matter for that.
If they hours didn't then it would be possible to have home-study for some time of the day and do all the bus transport mid 'school' day at various times for a massive reduction in resources.
I'm unclear as to why this change was made between one school year and another. Unless I'm mistaken, there was no lead time. This seems like a big bang software rewrite.
This kind of massive change should have a year or two or even four of prep work. People can adjust their jobs and schedules much more easily over that kind of time than a single summer.
It's true that it was fast, maybe too fast, although the district was planning this conversation for more than 10-20 years. But even if it has not been implemented, the debate itself may have been worth the trouble. The complex problem of start time choice is an incredibly deep and interesting question, and so is the interface between algorithms and public policy.
Sure, but 6 months is nothing. I'm talking about years. If parents knew that in 5 years the times would change, they could adjust their work schedules over that timeperiod. Especially if everyone in the whole district knew, even the childless.
There is still a single point where you have to go from, say, starting work at 9am vs starting work at 7am. And a lot of jobs do not provide this flexibility at all.
Can we please stop with writing out “... 50 bazillion zeros...” and just write using scientific notation? It’s part of a basic high school education and is much easier to read and understand.
>Even if the algorithm promised to reduce inequities, the upheaval involved — with nearly 85 percent of the district getting new start times — would hit black and brown families especially hard, the groups argued.
The article never really goes into this any farther. Was this just an empty point from the selfish wealthy families, or were they correct?
It makes sense that it can be easier to be misunderstood by a human than by a robot. And what about the case when any decision will always be bad for some people? I would argue that using (at least partly) an algorithm can add some fairness to the process, as without an algorithm systems can favor people who know the system the most.
Something like this just happened in the San Francisco school district. As of last month they decided to undo a key piece of an Ivy League authored school assignment algorithm and there's talk of scrapping the whole thing.[1]
If you know a parent with small children in SF you've probably have heard some consternation about the school lottery. Although the majority of families get their first choice, there are copious horror stories. Top economists were brought in around 2010 to design a new school assignment system.[2] They designed an algorithm. It tried to solve for a few goals: diversity, proximity and choice. They tried to make it "strategically simple" to avoid gaming the system.
The algorithm works like this: You rank the schools you want. Assignments are handed out according to a series of preferences called tiebreakers: First to siblings, then residents of low test score areas (to promote diversity), then residents of the neighborhood, then by lottery draw. If you didn't get anything on your list it'd assign you to the nearest school with space.
Then it did a neat thing: If it saw a mutually advantageous trade between families, it'd do it. If Family A got B's top choice and B got A's top choice they'd both get upgraded. It's called the Transfer Mechanism, commonly known as "swapping". This was meant to cut out complex strategies to get the school you want so you could just honestly list the schools you wanted in order.
Welp... After many years of horror stories of families getting assigned clear across town, the verdict finally arrived that the system utterly failed to improve diversity too. Here's what's going wrong:
1. That low test score area preference (called CTIP) is disproportionately used by the few white and asian families in those neighborhoods to get out of their local school. It's a golden ticket to any school in the city. Perversely, what was meant to help diversify actually increased segregation while rewarding the most privileged families.
2. The neighborhood preference (called Attendance Area) is woefully inadequate to ensure assignment to a nearby school. Many schools don't have enough capacity for their area, and if you don't get into your neighborhood school the other nearby schools might also be filled up by their own locals. That means you could be assigned clear across town to the nearest school with capacity -- which means not a very popular school either. No consideration is given to "bumping" a set of people over one neighborhood, for example, or otherwise more gently distributing the pain. (EDIT: I just noticed the original proposal had an "Overfill" preference to address this very issue.[2] Didn't get implemented.)
3. Far from simplifying strategy, the Transfer Mechanism actually opens up a whole new kind of strategy. Clever families list schools with "swap value" in the hopes that they'll transfer their way into an otherwise inaccessible school. This is exacerbated by the existence of speciality schools that don't have an Attendance Area preference, i.e. language immersions and K-8 programs, creating a kind of niche swapping opportunity with AA-bearing families. This results in excessive "false" applications to these speciality schools, bumping more "true" applicants to their oversubscribed local school, bumping other locals, replacing them with swapped-in clever families, and creating a sense of injustice. Worst of all, this strategy is, you guessed it, disproportionately utilized by privileged families.
So last month SFUSD just decided to suspend the Transfer Mechanism.[4] This is big news and I'm not sure most families realize it yet. And there is a proposal in the works to rethink the whole system, taking inspiration from Berkeley.[1]
There's a function which takes a schedule and generates a score. Leaving a spot unfilled or scheduling someone who isn't available subtracts lots of points. Scheduling someone much more or less often than the median subtracts points based on how much of a difference there is. Scheduling someone on a day they prefer adds points, and so forth.
Then it's a simple hill-climbing algorithm starting from an empty schedule. I get people's availability and preferences into the program, hit run, and out pops a schedule.
I learned early on that I must check the result before publishing it. The algorithm itself works fine. Hill climbing probably isn't finding the absolute optimal schedule, but it does a good job. The problem, of course, is the scoring function. Sometimes it turns out to be missing some important constraint. Much more often, I need to tinker with the weights assigned to various conditions.
People assign too much agency and intelligence to computers. They are, of course, just tools. A nail gun will put a nail into whatever you point it at, whether it's drywall or flooring or feet. Likewise, a computer optimization system will optimize whatever function you tell it to, and will probably do an excellent job, but it's up to you to make sure the function accurately describes what the humans want.