Common Lottery Algorithm

[Anonymous]

Anonymous wrote:

Anonymous wrote:hmmm...good point. I'm the PP who came up with the theory on how the lottery will work and you're right...it's not clear how the waitlist piece would then work...we'll it's clear how it would work for your top choice...but for the 11 other schools that you ranked lower, not so much. Yeah, I don't know. Again, this shouldn't be a guessing game. This is a lottery that our taxes fund...we should be allowed to understand the methodology.

my understanding (from a conversation with myschools) but I may have misunderstood, is that if you get into school #5 you will be waitlisted at schools #1-#4 and you will NOT be waitlisted at schools #6-#12

you're right. I was thinking backwards. Anyway, back to the original question...how would a student get waitlisted for all the schools ranked "higher" than a school where a student did receive a seat (or a student who didn't get any matches).

[Anonymous]

Found this more detailed description of the Denver system, which is designed by the same team and has been described as the model for the DC system:

http://www.aplusdenver.org/_docs/FINAL_TC%20Letter%20to%20Denver%20Community%20Members.pdf

Can't cut and paste properly, but the algorithm steps are described in page 4 of that pdf. Key things I took away

The lottery is run for each school, based on the ranking of the students of that school. So all the students who ranked that school #1 are run through the lottery and randomly assigned. Then the students who ranked it #2 are slotted in below them, etc.

This also helps:

http://media.nola.com/education_impact/photo/diagram-enrollment-041512jpg-aea0b995c0aa929b.jpg

That's the New Orleans system. It seems to me to imply a lottery for each school individually.

These are all designed according to the so-called "Deferred Acceptance Algorithm", which goes back to 1962. It's commonly accepted to be the best way to do this sort of thing.

[Anonymous]

my understanding (from a conversation with myschools) but I may have misunderstood, is that if you get into school #5 you will be waitlisted at schools #1-#4 and you will NOT be waitlisted at schools #6-#12

NP here. This is my understanding as well. I don't know if your ranking for a school gives you greater consideration or preference, however. It's a straight lottery for each school, with the same preferences (sibling, proximity), but this system aims to lower overall waitlist numbers by dropping you from every school that's lower in your ranking.

Adding a preference for your ranking would complicate things a bit too much, I believe.

[Anonymous]

my understanding (from a conversation with myschools) but I may have misunderstood, is that if you get into school #5 you will be waitlisted at schools #1-#4 and you will NOT be waitlisted at schools #6-#12

NP here. This is my understanding as well. I don't know if your ranking for a school gives you greater consideration or preference, however. It's a straight lottery for each school, with the same preferences (sibling, proximity), but this system aims to lower overall waitlist numbers by dropping you from every school that's lower in your ranking.

Adding a preference for your ranking would complicate things a bit too much, I believe.

[Anonymous]

Anonymous wrote:Found this more detailed description of the Denver system, which is designed by the same team and has been described as the model for the DC system:

http://www.aplusdenver.org/_docs/FINAL_TC%20Letter%20to%20Denver%20Community%20Members.pdf

Can't cut and paste properly, but the algorithm steps are described in page 4 of that pdf. Key things I took away

The lottery is run for each school, based on the ranking of the students of that school. So all the students who ranked that school #1 are run through the lottery and randomly assigned. Then the students who ranked it #2 are slotted in below them, etc.

This also helps:

http://media.nola.com/education_impact/photo/diagram-enrollment-041512jpg-aea0b995c0aa929b.jpg

That's the New Orleans system. It seems to me to imply a lottery for each school individually.

These are all designed according to the so-called "Deferred Acceptance Algorithm", which goes back to 1962. It's commonly accepted to be the best way to do this sort of thing.

Very interesting, thanks for posting! It seems if the Denver model is the case then there is a real incentive to strategically change the ordering. For example, if IT is my first choice school, but IT is not likely to have any non-sibling spots, then I might put another school first so that I have a shot at getting into that other school and don't waste my #1 slot on a school that will have few if any openings.

If the New Orleans model is the case, then the lottery will be run as in last year, and then the computer will make various trades to get people into their best options. So I put A first then B, but get into B; another person puts B before C but gets into C; and a third person puts C before A but gets into C. They get reordred so that everyone gets into their preferred choice. If this is the case, then it sounds great--like last year's, where it's all random, but people are more likely to get into a top choice.

Really hope that they give us more info soon.

[Anonymous]

Anonymous wrote:Found this more detailed description of the Denver system, which is designed by the same team and has been described as the model for the DC system:

http://www.aplusdenver.org/_docs/FINAL_TC%20Letter%20to%20Denver%20Community%20Members.pdf

Can't cut and paste properly, but the algorithm steps are described in page 4 of that pdf. Key things I took away

The lottery is run for each school, based on the ranking of the students of that school. So all the students who ranked that school #1 are run through the lottery and randomly assigned. Then the students who ranked it #2 are slotted in below them, etc.

This also helps:

http://media.nola.com/education_impact/photo/diagram-enrollment-041512jpg-aea0b995c0aa929b.jpg

That's the New Orleans system. It seems to me to imply a lottery for each school individually.

These are all designed according to the so-called "Deferred Acceptance Algorithm", which goes back to 1962. It's commonly accepted to be the best way to do this sort of thing.

Thanks for this. The first link is not likely as it doesn't fit with how they have described it. The New Orleans system seems more possible.

Does anyone know how the algorithm worked with the previous DCPS system only?

[Anonymous]

Anonymous wrote:It is the same method as last year, only with more schools. Think of it as mini-lotteries for each school, then a sorting process that gives you your highest ranked spot and puts you on the waiting list for any school you ranked higher.

The part you quote was also the same as last year (if you get in a school, you stay on the waiting list for those you ranked higher), but it the common lottery folks are saying it is NOT the same as last year. Something about the algorithm has changed, although I can't explain what.

[Anonymous]

Anonymous wrote:Found this more detailed description of the Denver system, which is designed by the same team and has been described as the model for the DC system:

http://www.aplusdenver.org/_docs/FINAL_TC%20Letter%20to%20Denver%20Community%20Members.pdf

Can't cut and paste properly, but the algorithm steps are described in page 4 of that pdf. Key things I took away

The lottery is run for each school, based on the ranking of the students of that school. So all the students who ranked that school #1 are run through the lottery and randomly assigned. Then the students who ranked it #2 are slotted in below them, etc.

This also helps:

http://media.nola.com/education_impact/photo/diagram-enrollment-041512jpg-aea0b995c0aa929b.jpg

That's the New Orleans system. It seems to me to imply a lottery for each school individually.

These are all designed according to the so-called "Deferred Acceptance Algorithm", which goes back to 1962. It's commonly accepted to be the best way to do this sort of thing.

As a followup, the system is being designed by these people:

http://www.iipsc.org/index.htm

It's chair is Alvin Roth, who won the Nobel Prize in economics for his work in this area. I have a hunch they know what they are doing.

http://www.nytimes.com/2012/10/16/business/economy/alvin-roth-and-lloyd-shapley-win-nobel-in-economic-science.html?pagewanted=all

[Anonymous]

Anonymous wrote:Found this more detailed description of the Denver system, which is designed by the same team and has been described as the model for the DC system:

http://www.aplusdenver.org/_docs/FINAL_TC%20Letter%20to%20Denver%20Community%20Members.pdf

Can't cut and paste properly, but the algorithm steps are described in page 4 of that pdf. Key things I took away

The lottery is run for each school, based on the ranking of the students of that school. So all the students who ranked that school #1 are run through the lottery and randomly assigned. Then the students who ranked it #2 are slotted in below them, etc.

This also helps:

http://media.nola.com/education_impact/photo/diagram-enrollment-041512jpg-aea0b995c0aa929b.jpg

That's the New Orleans system. It seems to me to imply a lottery for each school individually.

These are all designed according to the so-called "Deferred Acceptance Algorithm", which goes back to 1962. It's commonly accepted to be the best way to do this sort of thing.

THANK YOU!!! I am the poster in the other thread who kept saying it really does matter how you rank and if you rank a highly popular school #2, you are all but out of luck to get in if hundreds of people ranked it #1. That would be true under the Denver system, which I've also heard common lottery staff mention in talking about how this DC common lottery would run.

And to everyone saying the only differences this year are more schools and it includes the lottery, I'm with the other PP who said unless you identify yourself as an official spokesperson for the Common Lottery, you need to stop spreading misinformation. For those of us who have spoken to common lottery staff, we may not all be agreeing on what we heard, but one thing we all are agreeing on: the lottery itself will be different this year somehow and your ranking of schools matters more this year than it did last year. How? We're all still trying to agree on that. But it is NOT true that it's the same except for charters included and more schools.

[Anonymous]

Anonymous wrote:

Anonymous wrote:Found this more detailed description of the Denver system, which is designed by the same team and has been described as the model for the DC system:

http://www.aplusdenver.org/_docs/FINAL_TC%20Letter%20to%20Denver%20Community%20Members.pdf

Can't cut and paste properly, but the algorithm steps are described in page 4 of that pdf. Key things I took away

The lottery is run for each school, based on the ranking of the students of that school. So all the students who ranked that school #1 are run through the lottery and randomly assigned. Then the students who ranked it #2 are slotted in below them, etc.

This also helps:

http://media.nola.com/education_impact/photo/diagram-enrollment-041512jpg-aea0b995c0aa929b.jpg

That's the New Orleans system. It seems to me to imply a lottery for each school individually.

These are all designed according to the so-called "Deferred Acceptance Algorithm", which goes back to 1962. It's commonly accepted to be the best way to do this sort of thing.

As a followup, the system is being designed by these people:

http://www.iipsc.org/index.htm

It's chair is Alvin Roth, who won the Nobel Prize in economics for his work in this area. I have a hunch they know what they are doing.

http://www.nytimes.com/2012/10/16/business/economy/alvin-roth-and-lloyd-shapley-win-nobel-in-economic-science.html?pagewanted=all

Oh, I agree that the company designing the algorithm knows what it is doing. I just don't have faith that the DC representatives understand it or are able to communicate how it will work with the public, which is key to making your lottery selections.

And to the PP who insisted that this is the same lottery as last year, just with more schools, I really don't think that is correct.

[Anonymous]

Anonymous wrote:

my understanding (from a conversation with myschools) but I may have misunderstood, is that if you get into school #5 you will be waitlisted at schools #1-#4 and you will NOT be waitlisted at schools #6-#12

NP here. This is my understanding as well. I don't know if your ranking for a school gives you greater consideration or preference, however. It's a straight lottery for each school, with the same preferences (sibling, proximity), but this system aims to lower overall waitlist numbers by dropping you from every school that's lower in your ranking.

Adding a preference for your ranking would complicate things a bit too much, I believe.

No, the contrary is true in a system where there is far more demand for a smaller number of schools than there is slots. Weighting your ranking as a factor in how lottery numbers are assigned creates exactly the incentive that has been discussed for researching ahead of time and applying to your true first choice schools, because if it's not reallly your first choice (or not enough of a choice for you to risk getting in and then losing out on all the schools you ranked after it), why is it fair for you to get that slot over someone for whom it really IS their first choice?

And yes, I know there is much about this system that seems unfair. But with the limited number of lottery slots at the most popular schools, the only way you can say you're "improving the matching between families who really want the school and the school" and the only way it can matter how you rank the schools is if you give people who rank a particular school #1 an advantage in the lottery over everyone who ranked it 2 and lower. Then you move to thos who ranked it 2nd, and so on. That's the only way that "matching" and ranking can really have an impact in connecting a family who is prioritizing that school over the family who is ambivalent or doens't know what other school to put down.

[Anonymous]

Anonymous wrote:

Oh, I agree that the company designing the algorithm knows what it is doing. I just don't have faith that the DC representatives understand it or are able to communicate how it will work with the public, which is key to making your lottery selections.

Agreed about the communications problems. But, this stuff is pretty complicated, and has been evolving for over 50 years, so it's not exactly the easiest thing to boil down to a few bullet points. I do think a flow chart or diagram like the one from the New Orleans newspaper would be very helpful.

[Anonymous]

The main thing EVERY family doing the common lottery this year needs to understand is this:

Research your choices and make sure our #1 is really the school you would be happiest with if you got in. You ranking it #1 will affect both your chances of actually getting any available lottery slots, as well as your position on the waitlist. And rank in your true rank order.

Can we all at least agree that that is the approach families should take to ranking their 12 choices?

[Anonymous]

Anonymous wrote:

Anonymous wrote:

my understanding (from a conversation with myschools) but I may have misunderstood, is that if you get into school #5 you will be waitlisted at schools #1-#4 and you will NOT be waitlisted at schools #6-#12

NP here. This is my understanding as well. I don't know if your ranking for a school gives you greater consideration or preference, however. It's a straight lottery for each school, with the same preferences (sibling, proximity), but this system aims to lower overall waitlist numbers by dropping you from every school that's lower in your ranking.

Adding a preference for your ranking would complicate things a bit too much, I believe.

No, the contrary is true in a system where there is far more demand for a smaller number of schools than there is slots. Weighting your ranking as a factor in how lottery numbers are assigned creates exactly the incentive that has been discussed for researching ahead of time and applying to your true first choice schools, because if it's not reallly your first choice (or not enough of a choice for you to risk getting in and then losing out on all the schools you ranked after it), why is it fair for you to get that slot over someone for whom it really IS their first choice?And yes, I know there is much about this system that seems unfair. But with the limited number of lottery slots at the most popular schools, the only way you can say you're "improving the matching between families who really want the school and the school" and the only way it can matter how you rank the schools is if you give people who rank a particular school #1 an advantage in the lottery over everyone who ranked it 2 and lower. Then you move to thos who ranked it 2nd, and so on. That's the only way that "matching" and ranking can really have an impact in connecting a family who is prioritizing that school over the family who is ambivalent or doens't know what other school to put down.

I disagree with the bolded statement. I think if the Denver system is adopted, I'll have a real incentive to reorder my preferences to put schools in the 1-3 slots that are likely to have seats for non-IB, Non-sibs. Otherwise, I will be wasting my true #1 spot (for a popular charter, which will have few seats available).

[Anonymous]

Anonymous wrote:

Anonymous wrote:

my understanding (from a conversation with myschools) but I may have misunderstood, is that if you get into school #5 you will be waitlisted at schools #1-#4 and you will NOT be waitlisted at schools #6-#12

NP here. This is my understanding as well. I don't know if your ranking for a school gives you greater consideration or preference, however. It's a straight lottery for each school, with the same preferences (sibling, proximity), but this system aims to lower overall waitlist numbers by dropping you from every school that's lower in your ranking.

Adding a preference for your ranking would complicate things a bit too much, I believe.

No, the contrary is true in a system where there is far more demand for a smaller number of schools than there is slots. Weighting your ranking as a factor in how lottery numbers are assigned creates exactly the incentive that has been discussed for researching ahead of time and applying to your true first choice schools, because if it's not reallly your first choice (or not enough of a choice for you to risk getting in and then losing out on all the schools you ranked after it), why is it fair for you to get that slot over someone for whom it really IS their first choice?

And yes, I know there is much about this system that seems unfair. But with the limited number of lottery slots at the most popular schools, the only way you can say you're "improving the matching between families who really want the school and the school" and the only way it can matter how you rank the schools is if you give people who rank a particular school #1 an advantage in the lottery over everyone who ranked it 2 and lower. Then you move to thos who ranked it 2nd, and so on. That's the only way that "matching" and ranking can really have an impact in connecting a family who is prioritizing that school over the family who is ambivalent or doens't know what other school to put down.

Agreed wholeheartedly.