Problems with Rolling-Date First Come First Served

Before discussing ways to “fix” this system, I think it’s valuable to articulate several reasons that this approach isn’t very good. Often, people agree that a system is broken, but disagree on what the problem actually is (and therefore, on how to solve it). Here is my synthesis of various concerns I have heard expressed about first-come, first-served reservation systems. Although the concerns are many, I will place them into three groups.

1. The system is unpleasant.

• Having to arrange your day to be available at a particular time is inconvenient.
• Rushing when booking is stressful.
• Heavy traffic often causes servers to become overloaded, resulting in websites crashing, becoming unresponsive, losing your progress, and so on. This is frustrating.

2. The system is unfair.

• It favors people who can plan months in advance, can arrange their schedule to be available at a particular time, have a fast internet connection, and are technologically savvy.
• Some people might reserve multiple permits, while others are left with none.

3. The system is inefficient

• People must reserve permits before knowing their full schedule. If they choose not to use their permits and don’t cancel well in advance, this can result in vacancies, despite the high demand.

In 2025, according to US Forest Service data, 22,549 permits for the Boundary Waters Wilderness in Minnesota were booked on the first day reservations were open. Of these, 7,077 (31%) were eventually cancelled, and 2,177 (10%) resulted in a no-show without cancellation. Furthermore, approximately 1/3 of all cancellations happened within 4 days of the trip date.

• The one-day-at-a-time approach makes it difficult for people to get the dates they want.

Example 1. Family A wishes to travel in June but could possibly go in July, while family B is in the reverse situation. When June permits become available, family B applies for one, as they don’t know whether they will be able to get a permit for July. If family B is successful, then both families will also try to get a permit for July (A because they do not yet have a permit, and B because they prefer July). Either option could result in inefficiency: if A wins the July permit, both families get to camp, but would prefer to swap dates. If B wins the July permit, then they hold two permits, and the June permit might not be used at all (see first point above).²

• Loopholes in the system incentivize people to book permits they never intend to use. You can often book multiple consecutive nights at the same site, and only your start date must be within four months. This can be exploited, as illustrated by the following example.

Example 2. A family wants a campsite for June 4-5. On February 1, it logs in and books a five-night trip for June 1-5. The permits for June 1-3 are never used, and families that try to book for June 4 on February 4 find that these permits are already claimed.³

How I Would Change The System

There are many ways one could change this system, and there won’t be a single “best” one. Here, I describe three concepetual changes that I would support.

1. Move from daily release to seasonal release. Instead of permits for June 4 becoming available one day before permits for June 5, all permits for the summer season become available at the same time.

• Efficiency benefit #1: in Example 1 above, family B can directly reserve the July permit they want (leaving the June permit for someone else), rather than being forced to request the June permit before knowing whether the July permit is available.
• Efficiency benefit #2: this closes the “date sliding” loophold in Example 2. There is no longer an incentive to reserve more dates than you intend to use, as the permits for June 4 are made available at the same time, regardless of the start date of the trip.
• Precedent and Practicality: this is currently done for the Boundary Waters Wilderness in Minnesota, and for the 10th Mountain Division Huts in Colorado (see my previous blog post).

2. Lottery for access times, with advanced registration window. Instead of rewarding those who are available at 8 am and fast on their computers, have a long registration period (i.e. the month of January). Place everyone who registers during this time in a random order. Based on this order, assign each applicant an “earliest access time.” So for example, a lucky applicant might be told that they can reserve a permit starting on February 1 at 8 am, while other applicants might be told that they can log in on February 1 at 4 pm, or February 4 at 9 am.

• Fairness benefit #1: all applicants who sign up during the registration window are treated equally. There is no favoritism towards those who can be available at 8 am or those who are fast with computers (favoritism towards those capable of planning ahead remains).
• Fairness benefit #2: even if you aren’t able to log on exactly at the time you are assigned, the cost of logging in a few hours later is that a few more people have booked, rather than that all of the permits have been claimed.
• Pleasant experience benefit: you register at a time convenient to you, and when you do log on, you are not competing with many other people. The servers are not overloaded and do not crash.
• Precedent and practicality: this is similar to the system that the University of Minnesota uses for course registration. Students are given different times at which they can access the registration system, based on credits completed and a random lottery number. The US Government has had to move from FCFS to registration period + lottery when allocating both H1B and H2B visas. You can read about the change for H1B visas in this paper, and about the change for H2B visas in my previous blog post. In these cases, there was no need to create an access time: every applicant just wants the same thing (a visa). Note that this change really only makes sense if Change #1 has already been implemented: having separate registration windows for each date becomes cumbersome, and if all applicants must be assigned a time in a single day, then many applicants may necessarily have the same/similar access times, making the booking process more stressful.

3. Limit initial bookings, and punish cancelations and no-shows.

   The first two changes could make the process less stressful, more fair, and prevent certain types of inefficiency. But they do not address the unfairness of a single applicant claiming many permits while others get none, nor do they address the inefficiency from applicants who book permits that they don’t ultimately use.

   I propose that initially, applicants should only be able to reserve permits for a single trip (defined by a maximum number of consecutive days). Only once all applicant access windows have passed can anyone reserve a second trip. In fact, it may make sense to have a break of a couple of weeks between these times, so that those who missed the initial registration window can reserve an unclaimed permit before anyone can claim a second permit.

• Fairness benefit: nobody can claim a second permit until everyone has had a chance to claim a first.

• Efficiency benefit: this would likely reduce cancellations and no-shows, as I suspect that many of these come from people who reserve multiple dates in advance, and only later decide which of these to use.

  If cancellations and no-shows remain an issue, they can be addressed through punishment. Rather than monetary punishment (which again raises fairness concerns), I favor limiting the ability to make future bookings (i.e. later access times for those who have “behaved badly” in the past). There are many details to work out about how to do this effectively, but I won’t go into them here.

• Practicality: changes #2 and #3 both require a way to prevent individuals from creating multiple accounts (or at least to make this difficult). Otherwise, people will have an incentive to register multiple times (making it more likely to get an early access time), an ability to reserve multiple permits (one through each account), and the ability to cancel without punishment (by booking using a different account than the one that is being punished).

  Preventing users from creating multiple accounts isn’t trivial, but can be managed by asking for identifying information such as legal name and birthday, as well as a phone number (which seem to be a bit harder to duplicate than e-mail accounts), and implementing on-site verification to make sure that the person who shows up to claim a site is actually the one who booked the permit (this is good anyway, as a way to prevent scalping and resale).

Other Topics Discussed

In our conversation, we discussed several other topics, such as:

• Closing the “date sliding” loophole by preventing modifications to reservations that are booked on the first available day. I was skeptical, as this feels like a band-aid over a bad system, rather than a true redesign. More specifically, this could lead to inefficiency when people’s plans change and the reservation cannot be changed. Additionally, it requires a ranger to enforce the “if you don’t show up on the first night, you forfeit the remaining reservation” policy. In addition to being costly, I could imagine many rangers being lenient – after all, the camper will probably say they were waylaid unexpectedly, and nobody else is going to use the site on such short notice.
• Charging money for cancellations or no-shows. Charging for cancellations potentially reduces the number of bookings people make speculatively, but also reduces or eliminates their incentive to cancel when they know their plans have changed. The combined effect is not obvious. People are already charged for no-shows (in that they don’t get a refund), and I doubt there would be much appetite to levy an additional fine above the cost of the permit. This would also require careful monitoring of no-shows, which can be difficult in the backcountry. Finally, any sort of monetary punishment raises fairness concerns, in that some users can pay these costs much more readily than others. I generally prefer non-monetary punishment that restricts future bookings.
• Weighted lottery based on previous success/failure. This came up briefly in the article: the idea is to bias the lottery in favor of people who have applied in previous years and been unsuccessful. I mention in the article that this has been done for certain ultramarathons – for more details, see this blog post. This is also common practice for states that use lotteries to allocate hunting licenses – see for example North Dakota’s policies (and search for “Hunting License” “Preference Points” to learn more about other states). The city of Beijing also holds a lottery to allocate a limited supply of new license plates each year, and biases the lottery in favor of past losers. While this can mitigate unfairness by evening out luck across years, it also creates some perverse incentives to apply in years that you aren’t actually interested, just to accrue priority for future years in which you might be interested.