Each depth is written as a self-contained route. Choose one without needing to read the other two, or use Read all for a continuous article.
Intuition
A source does not disappear all at once
An astronomical signal can contain continuum power, spectral lines, timing structure, polarisation and spatial detail. Those parts do not have to fail together.
A source may remain detectable after some of the information that identified it has left the receiver band, fallen below the noise, been mixed into a background, or become too weak to distinguish from another possible history.
The edge changes what we see
Now add one more step. If weak and ordinary sources become unrecoverable first, the observable population near the limit will not look like a uniformly dimmed copy of the nearby universe.
It will become increasingly selected.
The horizon can therefore become visually strange before it becomes dark.
Why quasars belong here
Quasars are a natural object with which to ask this question because their enormous luminosities and distinctive spectra allow some of them to be identified at very large redshifts.
This does not mean that the soft horizon creates quasars or explains their internal physics. It means that an exceptionally luminous source is exactly the kind of signal expected to remain in a flux-limited and information-limited sample after less extreme sources have become difficult to recover.
The Argument
A complex source carries several kinds of information
For one emitted feature, a redshifted reception law gives
If the receiver is usable only above a lower boundary \(f_{min}\), that particular feature leaves the band when
A second feature at another emitted frequency reaches that boundary at another redshift. Information can therefore be lost progressively even while total received power remains non-zero.
Flux adds a second selection
Write the received flux schematically as
where \(\mathcal T\) collects the transmission effects assigned to the path and receiver. A survey with threshold \(F_{min}\) detects only signals satisfying
As distance increases, that condition increasingly favours intrinsically luminous sources, favourable spectra and signals whose identifying structure remains inside the usable observing window.
Detection and reconstruction are different thresholds
A source can pass the flux threshold and still fail the identification threshold. Conversely, a distinctive spectral pattern may allow identification at lower signal-to-noise than an otherwise featureless source.
The relevant horizon is therefore not one number. It is a family of boundaries depending on luminosity, spectrum, variability, polarisation, channel properties and instrument.
The population near the boundary is biased
Let \(\theta\) describe source properties and \(D\) the path. Then the observed population is weighted by a selection function
Far from the boundary, many kinds of source can have \(S\approx1\). Near the boundary, \(S\) can differ strongly between source classes.
Quasars are a useful stress case
High-redshift quasar surveys are explicitly flux limited, and deeper surveys recover fainter quasars than shallower ones. That ordinary observational fact already demonstrates how strongly the detected population can depend on the threshold.
Path I adds a further question: if the propagation channel also erodes discriminating information, does the effective selection become stronger than flux loss alone predicts?
Quasars do not answer that question by their existence. They provide an extreme class on which such a channel-selection prediction could be tested.
Deep Notes
The soft horizon is best treated as an information-and-selection problem, not as a new hard surface in space. The previous themes asked whether propagation can rescale an electromagnetic signal, transform its temporal structure and contribute to a diffuse background. Once those possibilities are allowed, the final question is what information about a source still survives after a very long path.
That immediately creates two different boundaries. The first is a detection boundary: does enough signal reach the instrument to register something at all? The second is a reconstruction boundary: does the surviving signal contain enough independent structure to decide what source and what path produced it?
Those boundaries need not coincide. More importantly, they need not be the same for every source. The horizon is therefore soft not only because information disappears gradually, but because different classes of object disappear from the reconstructed sky at different rates.
A feature-dependent spectral boundary
Take one emitted feature at \(f_{emit}\). If its received frequency is
and the observing system is useful only for \(f\ge f_{min}\), then that feature remains directly accessible only while
Define the corresponding feature horizon by
Another feature has another \(z_h\). A complex source can therefore lose one diagnostic while retaining another.
Bandwidth is only the simplest case
A real reconstruction can fail for many reasons before a feature literally leaves the observing band. Its amplitude may fall below noise. A line may be absorbed or blended. Temporal structure may become too slow, too weak or too poorly sampled. Polarisation may become uncertain. A diffuse background may dominate the contrast.
The relevant information is not total electromagnetic energy. It is discriminating structure: the part of the signal that rules out competing source histories.
Flux-limited selection
Even without any speculative channel effect, a survey has a selection function. In a schematic Euclidean form,
A detection threshold requires
At larger effective distance, a larger intrinsic luminosity or more favourable transmission is required. The observed sample therefore becomes increasingly biased toward sources able to remain above threshold.
Cosmological surveys use the appropriate luminosity-distance and K-correction machinery rather than this simple Euclidean expression. The selection principle is the point: the detected catalogue is filtered by the observing system and by the received spectrum.
Reconstruction adds another selection function
Let \(\theta\) denote intrinsic source parameters and \(C\) describe the channel. The probability that a source enters a reconstructed catalogue may be written schematically as
The second factor matters because a detected excess is not automatically a uniquely classified source.
If channel effects progressively erase particular diagnostics, then \(P(\mathrm{identify}\mid\mathrm{detect})\) can become strongly source dependent even before the signal disappears.
What remains detectable need not be typical
This produces the central consequence:
The distant sky need not approach darkness by showing every source class at the same relative abundance and simply reducing their brightness. One class may disappear from reliable reconstruction early, another later, while an unusually luminous or distinctive class continues to be identified.
The resulting population can look more exotic even if the underlying population changed much less dramatically.
Quasars as the natural example
Quasars are exceptionally luminous active galactic nuclei with recognisable spectral signatures, so they are an obvious class to remain visible in surveys probing very large redshifts.
The observational literature already makes the ordinary selection effect explicit. SDSS work at \(z\sim6\) constructed complete flux-limited quasar samples, and deeper imaging recovered quasars one or two magnitudes fainter than the luminous objects found in the main survey.
That fact should be used as a control, not as evidence for the speculative horizon. It shows that changing the threshold changes the apparent distant population.
What Path I adds to the ordinary selection effect
The channel hypothesis makes an additional prediction only if propagation does more than geometric dilution and known absorption. If the channel progressively removes identifying information, then two sources with the same received broadband flux could have different probabilities of surviving as reconstructable objects.
A useful test would therefore compare source classes, spectral diagnostics and observing bands at similar path lengths. A genuine channel-selection effect should predict which kinds of information disappear first and should do so with parameters already constrained by the redshift, supernova and CMB themes.
Without such a prediction, “soft horizon” remains a descriptive metaphor rather than a physical model.
The horizon can look like an emergence
This is the place where the quasar idea becomes interesting. From the observer's side, an exceptional population may appear to emerge near the edge of reconstruction because ordinary populations have already become difficult to recover.
Nothing new has to be created at the horizon. The apparent emergence can be produced by differential survival through the observational and propagation filters.
This is a selection hypothesis, not a claim that current quasar observations require it.
Instrumental, astrophysical and physical horizons must remain separate
If a new telescope, wider band or deeper exposure restores the missing population, the previous boundary was observational. If known absorption by intervening matter explains the loss, the boundary belongs to established astrophysical propagation. A stronger soft-horizon claim begins only where a proposed additional channel law predicts a residual selection that those effects do not explain.
That separation is essential if the idea is to become testable.
The final statement of Path I
The first part of the soft horizon is therefore informational: detectability can survive after unique reconstruction fails.
The second part is population-level: what remains reconstructable near that boundary need not be typical of what is physically present there.
Quasars are not offered as proof of the horizon. They are the clearest example of the kind of exceptional signal that makes the distinction observable.