Published 09 February, 2018; last updated 08 March, 2021

Altitude of objects attained by man-made means has seen six discontinuities of more than ten years of progress at previous rates since 1783, shown below.

This case study is part of AI Impacts’ discontinuous progress investigation.

We looked for records in height from the ground reached by any object via man-made technology.

‘Man-made technology’ is ambiguous, but we exclude for instance objects tied to birds and debris carried up by hurricanes. We include debris launched unintentionally via gunpowder explosion, and rocks launched via human arms.

We measure ‘altitude’ from the ground at the launch site. This excludes mountain climbing, but also early flight attempts that involve jumping from towers and traveling downward slowly.¹⁾ It also excludes early parachutes, which were mentioned in fiction thousands of years ago.²⁾

Measured finely enough, there are never discontinuities in altitude, since objects travel continuously.³⁾ This prohibits finding discontinuities in continuously measured altitude, but doesn’t interfere with the dataset being relevant evidence to us. We are interested in discontinuities because they tell us about how much surprising progress can happen in a short time, and how much progress can come from a single innovation. So to make use of this data, we need to find alternate ways of measuring it that fulfill these purposes.

For the purpose of knowing about progress in short periods, we can choose a short period of interest, and measure jumps in progress made at that scale. For the purpose of knowing about progress made by single innovations, we can assign the maximum altitude reached to the time that the relevant innovation was made, for instance.⁴⁾

We could measure both of these trends, but currently only measure a version of the former. For short periods of travel, we assign the maximum altitude reached to the date given (our understanding is that most of the entries took place over less than one day). For travel that appears to have taken more than a day, we record any altitudes we have particular information about, and otherwise estimate records on roughly an annual basis, including a record for the peak altitude (and possibly more than a year apart to allow for the final record to have the maximum altitude). This is ad hoc, but for the current purpose, converting what we have to a more consistent standard does not seem worth it. Instead, we consider these the effects of these choices when measuring discontinuities. They do not appear to matter, except to make modest differences to the size of the pellet discontinuity, discussed below (section, ‘Discontinuity measurement’).

We collected data from various sources, and added them to this spreadsheet, tab ‘Manned and unmanned’. This data is shown in Figures 1-3 below. We have not thoroughly verified this data.

Record altitudes might plausibly be reached by a diversity of objects for a diversity of purposes, so collecting such data is especially dependent on imagination for the landscape of these.⁵⁾ For this reason, this data is especially likely to be incomplete.

We also intentionally left the data less complete than usual in places where completeness seemed costly and unlikely to affect conclusions about discontinuities. The following section discusses our collection of data for different periods in history and details of our reasoning about it.

Here we describe the history of progress in altitude reached and the nature of the data we collected during different times. See the spreadsheet for all uncited sources.

Chimps throw rocks, so we infer that humans have probably also done this from the beginning.⁶⁾ A good rock throw can apparently reach around 25m. Between then and the late 1700s, humanity developed archery, sky lanterns, kites, gunpowder, other projectile weapons, rockets, and primitive wings,⁷⁾ among probably other things. However records before the late 1700s are hard or impossible to find, so we do not begin the search for discontinuities until a slew of hot air balloon records beginning in 1783s. We collected some earlier records in order to have a rough trend to compare later advances to, but we are likely missing many entries, and the entries we have are quite uncertain. (It is more important to have relatively complete data for measuring discontinuities than it is for estimating a trend.)

The highest altitude probably attained before the late 1700s that we know of was reached by debris in a large gunpowder building explosion in 1280, which we estimate traveled around 2.5km into the air. Whether to treat this as a 'man-made technology' is ambiguous, given that it was not intentional, but we choose to ignore intention.⁸⁾

Kites may also have traveled quite high, quite early. It appears that they have been around for at least two thousand years.⁹⁾ and were used in ancient warfare and even occasionally for lifting people. We find it hard to rule out the possibility that early kites could travel one or two thousand meters into the air: modern kites frequently fly at 2km altitudes, silk has been available for thousands of years, and modern silk at least appears to be about as strong as nylon.¹⁰⁾ Thus if we are wrong about the gunpowder factory explosion, it is still plausible that two thousand meter altitudes were achieved by kites.

Over a period of three and a half months from August 1783, manned hot air balloons were invented,¹¹⁾ and taken from an initial maximum altitude of 24m up to a maximum altitude of 2700m. While this was important progress in manned travel,¹²⁾ most of these hot air balloons were still lower than the gunpowder explosion and perhaps kites. Nonetheless, there are enough records from around this time, that we begin our search for discontinuities here.

The first time that humanity sent any object clearly higher than ancient kites or explosion debris was December 1783, when the first hydrogen balloon flight ascended to 2,700m. This was not much more than we (very roughly) estimate that those earlier objects traveled. However the hot air balloon trend continued its steep incline, and in 1784 a balloon reached 4000m, which is over a thousand years of discontinuity given our estimates (if we estimated the rate of progress as an order of magnitude higher or lower, the discontinuity would remain large, so the uncertainties involved are not critical.)

The next hot air balloon that we have records for ascended nearly twice as high—7280m—in 1803, representing another over a thousand years of discontinuity. We did not thoroughly search for records between these times. However if that progress actually accrued incrementally over the twenty years between these records, then still every year would have seen an extra 85 years of progress at the previous rate, so there must have been at least one year that saw at least that much progress, and it seems likely that in fact at least one year saw over a hundred years of progress. Thus there was very likely a large discontinuity at that time, regardless of the trend between 1784 and 1803.

We collected all entries from Wikipedia’s Flight altitude record page, which claims to cover 'highest aeronautical flights conducted in the atmosphere, set since the age of ballooning'.¹³⁾ It is not entirely clear to us what 'aeronautical flights' covers, but seemingly at least hot air balloons and planes. The list includes some unmanned balloons, but it isn’t clear whether they are claiming to cover all of them. They also include two cannon projectiles, but not 38 cm SK L/45 "Max", which appears to be a record relative to anything they have, and cannon projectiles are probably not 'flights', so we think they are not claiming to have exhaustively covered those. Thus between the late 1700s, and the first flights beyond the atmosphere, the main things this data seems likely to be missing is military projectiles, and any other non-flight atmospheric-level objects.

We searched separately for military projectiles during this period. Wikipedia claims, without citation, that the 1918 Paris gun represented the greatest height reached by a human-made projectile until the first successful V-2 flight test in October 1942,¹⁴⁾ which matches what we could find. We searched for military records prior to the Paris gun, and found only one other, “Max” mentioned above, a 38cm German naval gun from 1914.

We expect there are no much higher military records we are missing during this time but that we could easily have missed some similar ones. As shown in Figure 1, the trend of military records we are aware of is fairly linear, and that line is substantially below the balloon record trend until around 1900. So it would be surprising if there were earlier military records that beat balloon records, and less surprising if we were missing something between 1900 and 1918. It seems unlikely however that we could have missed enough data that the Paris Gun did not represent at least a moderate discontinuity.¹⁵⁾

We could not think of other types of objects that might have gone higher than aeronautical flights and military projectiles between the record 1803 balloon and V-2 rockets reaching 'the edge of space' from 1942. Thus the data in this period seems likely to be relatively complete, or primarily missing less important military projectiles.

The German V-2 rockets are considered the first man-made objects to travel to space (though the modern definition of space is higher)¹⁶⁾ so they are presumably the highest thing at that time (1942). They are also considered the first projectile record since the Paris gun, supporting this. Wikipedia has an extensive |list of V-2 test launches and their outcomes, from which we infer than three of them represent altitude records.¹⁷⁾

The two gun records we know of were both German WWI guns, and the V2 rockets that followed were German WWII weapons, apparently developed in an attempt to replace the Paris Gun when it was banned under the Versailles Treaty.¹⁸⁾ So all altitude records between the balloons of the 1800s and the space rockets of the 50s appear to be German military efforts.

Between the last record V-2 rocket in 1946 and 1957, we found a series of rockets that traveled to increasing altitudes. We are not confident that there were no other record rocket altitudes in this time. However the rockets we know of appear to have been important ones, so it seems unlikely that other rockets at the time were radically more powerful, and there does not appear to have been surprising progress over that entire period considered together, so there could not have been much surprising progress in any particular year of it, unless the final record should be substantially higher than we think. We are quite unsure about the final record (the R-7 Semyorka), however it doesn’t seem as though it could have gone higher than 3000km, which would only add a further four years of surprising progress to be distributed in the period.

In October 1957, at least one centimeter-sized pellet was apparently launched into solar orbit, using shaped charges and a rocket. As far as we know, this was the first time an object escaped Earth’s gravity to orbit the sun.¹⁹⁾ This episode does not appear to be mentioned often, but we haven’t found anyone disputing its being the first time a man-made object entered solar orbit, or offering an alternate object

Because the pellets launched were just pellets, with no sophisticated monitoring equipment, it is harder to know what orbit they ended up in, and therefore exactly how long it took to reach their furthest distance from Earth, or what it was. Based on their speed and direction, we estimate they should still have been moving at around 10km/s as they escaped Earth’s gravity. Within a day we estimate that they should have traveled more than six hundred times further away than anything earlier that we know of. Then conservatively they should have reached the other side of the sun, at a distance from it comparable to that of Earth, in around 1.5 years. However this is all quite uncertain.

At around this time, reaching maximum altitudes goes from taking on the order of days to on the order of years. As discussed at the start of section 'Altitude of objects attained by manmade means' above, from here on we record new altitudes every year or so for objects traveling at increasing altitudes over more than a year.

In the years between 1959 and 1973, various objects entered heliocentric orbit.²⁰⁾ It is possible that some of them reached greater altitudes than the pellets, via being in different orbits around the sun. Calculating records here is difficult, because reaching maximal distance from Earth takes years,²¹⁾ and how far an object is from Earth at any time depends on how their (eccentric) orbits relate to Earth’s, in 3D space. Often, the relevant information isn’t available.

Among artificial objects in heliocentric orbit listed by Wikipedia²²⁾ none are listed as having orbits where they travel more than 1.6 times further from the Sun than Earth does,²³⁾ though many are missing such data. This is probably less far than the pellets, though further away than our conservative estimate for the pellets. For an object to reach this maximal distance from the Earth, it would need to be at this furthest part of its orbit, while being on the opposite side of the Sun from Earth, on the same plane as Earth.

Given all of this, it seems implausible that anything went ten times as far from the Sun as Earth by 1960, but even this would not have represented a discontinuity of even ten years. Given this and the difficulty of calculating records, we haven’t investigated this period of solar orbiters thoroughly.

In 1973 Pioneer 10 became the first of five space probes to begin a journey outside the solar system. In 1998 it was overtaken by Voyager 1. We know that no other probes were the furthest object during that time, however have not checked whether various other objects exiting the solar system (largely stages of multi-stage rockets that launched the aforementioned probes) might have gone further.

Figure 1 shows all of the altitude data we collected, including entries that turned out not to be records. Figures 2 and 3 show the best current altitude record over time.

For measuring discontinuities, we treat the past trend at a given point as linear or exponential and as starting from earlier or later dates depending on what fits well at that time.²⁴⁾ Relative to these previous rates, this altitude trend contains six discontinuities of greater than ten years, with four of them being greater than 100 years:²⁵⁾

The 1957 pellets would be a 66 year discontinuity if we counted all of their ultimate estimated altitude as one jump on the day after their launch, so exactly how one decides to treat altitudes that grow over years is unlikely to prevent these pellets representing a discontinuity of between ten and a hundred years.

In addition to the size of these discontinuities in years, we have tabulated a number of other potentially relevant metrics here.²⁶⁾

Primary authors: Katja Grace, Rick Korzekwa

Thanks to Stephen Jordan and others for suggesting a potential discontinuity in altitude records.