Channeling and Uneven Extraction

Phenomenon

You have dialed your espresso in to twenty-eight seconds, with exactly the same recipe as yesterday. But today's shot is sour and astringent at once: sharp acidity at the front of the tongue, and a dry mouth after you swallow.

You knock the spent puck out of the portafilter and find a few small pits on the surface and a thin crack down the side. If you were using a bottomless portafilter, you would also have seen this: as the shot ran, one spot on the bottom of the basket turned pale first, and a thin jet sprayed out sideways from there while the area around it was still dark.

The time was right — so why is the taste wrong? And what are those pits and that crack telling you?

Variables

The star of this chapter is not "how much was extracted" but "how evenly it was extracted". The main things that affect it:

And one reading: the difference from shot to shot. With the same recipe, a time that jumps between twenty-two and thirty-five seconds is a clue in itself.

Experiment

A thought experiment: imagine a puck as a left half and a right half. Both halves hold the same amount of coffee at the same fineness; only the left half is packed a little looser.

Write down your prediction before reading on

Which side will more of the water go through? As the shot goes on, will the difference grow, shrink or stay the same? How will the two halves end up extracted?

Your prediction

Save prediction

Stored only in this browser on this device. Never uploaded.

A second thought experiment: this "half loose, half tight" espresso happens to run in exactly the usual twenty-eight seconds.

Write down your prediction before reading on

From the time alone, could you spot the problem? What would this shot taste like?

Your prediction

Save prediction

Stored only in this browser on this device. Never uploaded.

Observation

In the first experiment, more of the water goes through the looser half. Under the same pressure, more of the flow takes the paths with less resistance Confidence: [Established Mechanism] — just as traffic crowds onto the clearer road.

The difference may even grow. Researchers have proposed a mechanism with a mathematical model: where more water flows, solubles are carried away faster, the coffee there becomes "emptier" and resists less, so even more water flows through — the shortcut reinforces itself Confidence: [Open Question]. This explanation so far comes mainly from modeling, but it fits what experiments have seen: past a certain fineness, espresso extraction yield actually drops and shots vary more from one to the next, which researchers attribute to water flowing unevenly through the puck Confidence: [Strong Evidence].

The result is that one puck holds two kinds of coffee at once: the part flushed by lots of water is over-extracted, and the part the water barely touched is under-extracted. A cup that is sour, bitter and astringent all at once is often a mix like this Confidence: [Rule of Thumb]: the overdone part brings bitterness and astringency, the underdone part brings sharp sourness.

The answer to the second experiment is: no. Total time only tells you the average resistance of the whole puck. One side fast, one side slow, and the average can land right on twenty-eight seconds. So a shot with the right time and the wrong taste is no contradiction.

Model

Water only cares about resistance

Picture the puck as a network of countless tiny channels. Pushed by pressure, water moves down through them, and the flow in each channel depends on the pressure pushing it and that channel's resistance Confidence: [Established Mechanism]. Wherever the puck is a little looser, a little thinner, or cracked, that is a low-resistance path.

Channeling means the water has concentrated in a few such shortcuts. In extreme cases you can see what it left behind on the spent puck: small holes, cracks, a wet ring around the edge.

Unevenness is the norm; the question is how much

Perfectly even extraction does not exist: particle size is a distribution to begin with (Chapter 20), and the density of a puck can never be identical everywhere. In the Atlas Extraction Bed, you can see that within one dose of coffee, different places are extracted to different degrees as a matter of course.

So the question is not "is there unevenness?" but how uneven is it? Slight unevenness cannot be tasted; once it becomes clear channeling, the taste veers in two directions at once, and every shot is different.

Espresso is more prone to this than pour-over: finer coffee, more resistance, higher pressure, shorter time. A tiny difference in density can be amplified into a shortcut.

What puck prep is defending against

Distribution and tamping have one goal: make the resistance the same everywhere in the puck, leaving the water no shortcuts.

• Distribute evenly. Ground coffee often lands in the basket as a little mound, with clumps. Many baristas stir the coffee with fine needles to break up clumps, then level the surface Confidence: [Common Practice]. The practice is common, but so far we have not found a peer-reviewed study that specifically tests how much it helps Confidence: [Open Question].
• Tamp level, not hard. Whether the tamp is level matters more than how hard it is Confidence: [Rule of Thumb]: a tilted tamp makes the puck thick on one side and thin on the other, and water heads for the thin side. As for force, one study found no detectable effect on extraction across the tamping pressures it tested Confidence: [Strong Evidence].
• The structure of the puck matters too. One study layered coarser and finer coffee in the basket and found that this made the flow rate more consistent from shot to shot Confidence: [Strong Evidence]. It shows that the same coffee, arranged differently, changes how water flows.
• Start gently. Many machines first wet the puck at lower pressure for a few seconds before going to full pressure; this is called pre-infusion Confidence: [Common Practice]. The idea is to let the puck absorb water and swell evenly first, reducing the chance of a shortcut being blasted open right at the start.

Another road: do not grind so fine

Another approach is not to make the puck more perfect, but to make shortcuts less likely in the first place. Research found that using a little less coffee and grinding a little coarser makes extraction more consistent from shot to shot Confidence: [Strong Evidence] (Chapter 32). A coarser grind resists less, so the same difference in density produces a less extreme difference in flow. The cost is a faster, weaker shot with a different flavor.

Pour-over has channels too

Channeling is not unique to espresso. Every method where water flows through a layer of coffee — percolation (Chapter 27) — runs into it: always pouring onto the same spot digs a crater in the bed; coffee stuck high on the filter walls lets water run down the walls and barely pass through any coffee. Immersion (Chapter 28) is less sensitive to this Confidence: [Rule of Thumb]: all the coffee sits in all the water the whole time, so there is no question of which path is easier.

Transfer

Next time a cup tastes "off somehow", use this table to look for clues. These are signs to suspect uneven flow, not a diagnosis Confidence: [Rule of Thumb]:

Fix evenness first, then adjust extraction. When evenness is the problem, adjusting grind by Chapter 32's table often just makes things more confusing.

Misconceptions

"The harder you tamp, the better the espresso." One study found no clear effect of the force itself on extraction across the range it tested. What matters more is tamping level and the same way every time. And if you tamp hundreds of times a day, pressing too hard is a burden on your wrist.

"If the time is right, the extraction is even." Time reflects only the average resistance of the whole puck. One side fast and one side slow can average out to a "standard" time while the cup holds two extractions mixed together.

"Channeling is only an espresso problem." Every percolation method meets it. Channels in pour-over are usually less violent, but concentrated pouring, an uneven bed and coffee stuck on the walls all leave part of the coffee barely extracted.

"With enough tools and careful enough technique, extraction can be perfectly even." Unevenness is the norm; tools can only reduce it. For many distribution tools and techniques, controlled studies are still missing. Rather than doing every step, do the same steps the same way each time, and check whether your results become more consistent.

Real Lab

This chapter has no real-lab card. Pick whichever of these fits the equipment you have.

With pour-over — no espresso machine needed. Change only one variable: how the water lands on the coffee.

1. Same coffee, same grind, same dose and water: brew two cups.
2. First cup: aim every pour at the dead center, fast and hard, with no stirring at all.
3. Second cup: pour slowly in circles so the whole bed is wetted evenly, and give the dripper a gentle swirl at the end so the bed settles flat.
4. Note both drawdown times; after brewing, look at the beds: which one has a crater in the middle, or coffee stuck on the walls?
5. Let both cool to about the same temperature and taste side by side: which is sweeter? Which has sourness and astringency at the same time?

With a moka pot. Moka coffee is not tamped, but distribution still matters. Do it twice: once tipping the coffee into the basket carelessly, higher on one side; once tapping it level and scraping it flat. Keep everything else the same, and compare whether the coffee comes out steadily, and how the taste differs.

With an espresso machine. Pull shots with the same grind and dose: one set where the coffee goes straight from the grinder to the tamp, one set where you break up clumps and level it first. Do two or three of each and see which set has more consistent times; with a bottomless portafilter, watch whether the flow appears evenly across the whole bottom, or sprays out of one spot first.

Write it down: what you changed, whether the time was consistent, and whether the taste veered in two directions at once.

Check Your Model

Predict. Your espresso is already on the slow side, thirty-two seconds, a little bitter. You want "a bit more extraction", so you grind one more step finer. Before looking at any result, write down: how will the time and taste change? Could something happen that you did not expect?

Diagnose. A friend says: "My recipe is the same every day, but the time is sometimes twenty-two seconds and sometimes thirty-five, and the taste is good one day and bad the next. Is my grinder broken?" Apart from the grinder, what would you ask them to check first?

Explain. In your own words, explain to a friend: why does water "find shortcuts" in the puck? Why does a badly channeled espresso taste sour and bitter at the same time?