Starlink review: broadband dreams fall to Earth

The problem with reality is that it is very irritating.

In our imaginations, we might dream of a satellite internet system that delivers lightning-fast broadband speeds from space, freeing us from the dreary earthbound experience of cable monopolies and wireless data caps. We might envision an ISP that smashes through the plodding local politics of digging fiber trenches by literally achieving escape velocity and delivering fast, reliable internet from the heavens above. A system that will work on moving trucks, RVs, and even boats. Space-based internet access that will change everything because there is nothing technology cannot achieve in our minds.

Starlink, a new satellite internet service from SpaceX, is a spectacular technical achievement that might one day do all of these things. But right now it is also very much a beta product that is unreliable, inconsistent, and foiled by even the merest suggestion of trees. 

Reality, it must be emphasized, is very irritating.

The Verge has not written a story about broadband access or telecom policy in recent memory without a chorus of commenters responding that Starlink would fix it in some way. Access gap? Starlink. Data caps? Starlink. Wackadoo net neutrality bullshit? Starlink will fix it.

I completely understand where this is coming from: most Americans live under the shadow of regional cable monopolies that dominate the broadband market with high prices, sadistic data policies, and poor customer support. If you do have the choice of a second internet provider, it is often based on older technology and offers much slower speeds.

In rural America, the situation is even worse: a combination of bad policy and greed means there are huge swaths of the country where people aren’t even connecting at 25Mbps down, the pathetic standard for “broadband speed” used by the federal government. Where I live in rural New York state, the best available data suggests that only 43 percent of people connect at 25Mbps or above. That is ridiculous, especially since wireless carriers, in particular, have begged for lax regulatory oversight against the promise of delivering rural broadband over LTE and now 5G. They have mostly failed to do so.

The idea of ordering a $499 dish with a $99 monthly fee that can deliver Starlink’s current goal of 100Mbps down and 20Mbps up would indeed be a dream come true — especially since Starlink has set a long-term goal of 1Gbps down. It represents competition, something the American broadband market sorely lacks.

In that context, Starlink also represents something else: the American telecom policy establishment’s long-standing, almost religious belief that consumers are best served by something called “facility-based competition.” Starlink is a new facility for accessing the internet, one that does not rely on existing infrastructure. “Facility-based competition,” telecom lobbyists feverishly whisper while handing out their dirty, sweat-stained checks in Congress. “That is the American way.”

Of course, the only thing a decades-long commitment to “facility-based competition” has brought to most Americans is… a total lack of competition. Reality, as I have said, is quite irritating.

(By contrast, in Europe, where the prevailing philosophy is called “service-based competition,” large incumbent providers are required to lease fiber access to competitors and there is a thriving market for internet access with much lower prices for much faster speeds. If the United States were in Europe, it would have the most expensive broadband in the region.)

Anyway, American broadband policy stinks, and we all pay too much money for slow speeds and terrible customer service. It is no wonder people are delirious with excitement about Starlink, which promises to provide access from a constellation of thousands of tiny satellites blanketing the Earth, using a cutting-edge phased array antenna in the dish to quickly track the satellites moving across the sky. When it is fully deployed, Starlink claims it will operate the world’s largest satellite constellation, managed by a new automated orbital guidance system and an automated collision avoidance system, which has already been involved in a controversial close call. (Starlink has also run into communications problems with other satellite operators.)

Starlink is a lot of very bold engineering advancements packaged up in a $499 consumer product; the whole thing is far more advanced than previous satellite internet systems, which are slow, heavily data-capped, and very expensive. 

The Starlink coverage map divides the globe into a honeycomb-like hexagonal grid; the satellites launched so far mostly provide service in the northern part of North America. The whole thing is still in beta, so access is limited — even if you’re in a coverage area, there are only so many available slots in each part of the grid, so as not to stress the system. 

Luckily, my area has not yet filled its allotment yet, so I was able to simply sign up on the website, and my kit arrived about a week later. Let’s smash the system, I thought.

Then I learned about trees.

Inside the large gray Starlink box, you’ll find four items: the dish itself, which is connected to a 100-foot power-over-Ethernet (PoE) cable; a short black metal tripod stand for the dish; the main black Starlink power adapter; and a small silver Wi-Fi router with its own white PoE cable. The fundamental setup is incredibly simple: you plug both Ethernet cables into the power adapter, plug that into the wall, and you’re done. The printed instructions in the box are just pictograms, like Ikea for space internet.

All of the hardware is nicely designed — even though it’s in beta, it feels close to a consumer product already, with a sense of style that goes well beyond the hospital equipment vibes of most satellite gear. The dish itself (officially named “Dishy McFlatface”) is made of white plastic, with a matte white texture on its face. Two buttons on the mounting pole click into the included tripod mount, and that’s that. There are motors that rotate and tilt the dish to align it automatically; no fiddling required.

(One design oversight: the cable is permanently attached to the dish, so if it gets damaged — it’s outside, after all — you’re likely looking at replacing the entire dish, not just the cable.)

Although the Starlink kit ships with a short tripod and the sparse online instructions refer to it being “knee-high,” the dish really needs to be mounted as high up as you can get it. Starlink requires near-perfect line of sight to its satellites, which are often fairly low in the sky. Trees, buildings, and even poles will easily obstruct the signal, so if you’ve got tall trees blocking the horizon there’s really no choice but to get up and over them. Starlink beta testers have gone to hilarious and wonderful DIY lengths to solve this problem. (If there is one unreservedly excellent thing about Starlink, it is the community of beta testers, who are all the sort of clear-eyed we’ll-figure-it-out nerds that lend early tech products an air of infectious discovery and enthusiasm. I love you, Starlink people.)

I am going to emphasize the line-of-sight requirement, since it is crucial to understanding what Starlink can and cannot do right now, and it’s an important reality check on what it might be able to do in the future. Like the similarly over-hyped mmWave 5G, Starlink is remarkably delicate. Even a single tree blocking the dish’s line of sight to the horizon will degrade and interrupt your Starlink signal. Whatever satellite internet dreams you may have will run crashing into this reality until you can literally rise above.

Starlink’s website makes all of this crystal clear. “If any object such as a tree, chimney, pole, etc. interrupts the path of the beam, even briefly, your internet service will be interrupted,” says Starlink. “The best guidance we can give is to install your Starlink at the highest elevation possible where it is safe to do so, with a clear view of the sky. Users who live in areas with lots of tall trees, buildings, etc. may not be good candidates for early use of Starlink.” (I encourage you to square the advice to mount the dish as high as possible with the Starlink team’s further recommendation to bring ol’ Dishy inside in high wind conditions. Keep that ladder handy.)

Why am I hammering this point home? Because Starlink’s solution to the line-of-sight issue is to put more satellites into space, and, well, that’s not necessarily great. While Starlink has an army of devoted heart-eyed fans, it has an equal number of critics in the scientific community who note that blanketing the sky with tiny satellites will interfere with astronomers the world over. Starlink satellites are already bright enough to confuse people, and their potential to interfere with telescopes is well-documented. (No, you cannot just paint them black because the idea is to look at space, not thousands of little black satellites.)

It is a pretty damning indictment of broadband policy in the United States that a lot of people are so desperate for competitive options that they’re like “fuck telescopes.” But here we are.

Anyway, once you’ve got everything set up, the Starlink dish sweeps across the sky keeping track with the satellites, so finding a single open spot isn’t enough. You need a wide swath of open sky — Starlink suggests a cone of about 100 degrees, with a minimum elevation of 25 degrees above the horizon. So: low and wide. The app will tell you where and how the view to the satellite is obstructed, by how much, and how many hours a day it will affect your signal. I have my dish 60 feet away from my house with clear views of the sky, and it is still obstructed for two hours a day because of the very top of my house and the trees behind it. If this wasn’t a short-term review, I would certainly have it mounted on a pole on top of the structure.

Once you’ve got your dish in a good spot and figured out how to get the cable into your house (another moment when having it permanently attached to the dish is less than ideal, since this will generally require drilling and snaking the cable through a wall), you plug it all in and wait. The dish will rotate around to find a signal and then download the satellite schedule so it can keep itself aligned.

While that’s happening, the Wi-Fi router will boot itself up (it’s a wee bit slow) and eventually offer you a generic Starlink network. Connect to that, open the Starlink app, and you’ll be given the chance to rename that network and set a Wi-Fi password. And that’s it. It is remarkably simple and easy for a deeply advanced satellite internet system; there is nothing else to configure or otherwise worry about.

The included dual-band Wi-Fi 5 router is extremely bare-bones: it has one additional Ethernet port that can support a switch and exactly no listed specifications or software options at all. If you don’t want to use Starlink’s Wi-Fi router, you don’t have to — if your preferred Wi-Fi router supports static routes, you can plug it into the Starlink power box and apparently use that instead. (I did not test this because my partner is a literal divorce lawyer, and I was not eager to tell her the Wi-Fi network she relies upon for work would be intermittently brought to a halt by the presence of trees.)

Once you’re all set up and plugged in, there’s not much to say. Starlink offers a moderately fast, very inconsistent broadband connection. I definitely saw speeds that exceeded the promised 100Mbps down, topping out at 222Mbps down and 24Mbps up. But my usual speeds hovered between 30 and 90 down, matching what others have reported, and the connection slowed down and dropped out with surprising frequency.

If Starlink could offer consistently fast speeds, it would be competitive with the fastest package I can get from my rural cable provider, which tops out at $200 / mo for 325 / 25 but is still not attractively priced compared to the services available in more populated areas. 

In my week of testing, Starlink was perfectly fine for anything that buffers — I was able to stream Netflix and Disney Plus in 4K and jump around YouTube videos without significant issues — but doing something faster-paced, like quickly scrolling through TikTok videos, would run into delays.

Services that require a sustained, real-time connection, like Slack, Zoom, or gaming, simply weren’t usable for me, even when I was seeing the fastest speeds. I had high hopes that I could spend several days working over Starlink, and after just a few lost Slack messages and Zoom calls where my video dropped to low resolution and then froze entirely, I gave up. Many Starlink beta testers similar report experiences — consistent dropouts of a few seconds, every few minutes.

Starlink’s latency also swings from fine — Zoom did not exhibit any delay when it worked — to pretty bad. My feeling is that the connection dropouts are going to be worse for gaming than latency, so I didn’t spend any time testing gaming latency, but Starlink itself measures ping times for Counter-Strike: Go and Fortnite in its app, and I rarely saw those numbers dip below 50ms, mostly hovering around 85-115ms. Those aren’t numbers you’d want to game with, unless you like losing. (Some Starlink testers have been able to play games and even use Stadia, but that seems both inconsistent and heavily dependent on satellite coverage in your area.)

There are no data caps right now, but Starlink is clearly thinking about it, using the same “preventing abuse” language as any other broadband provider. If you are dreaming of signing up for Starlink as a way to tell your local cable monopoly to kick rocks, well, consider what might happen when Starlink is your space-based internet access monopoly.

Look, I know you’re hyped up about Starlink. I feel you. I also wish I could tweet a photo of Dishy in my yard to every telecom CEO in the game and tell them to try harder. But The Verge has long had a hard rule against reviewing products based on potential because the sad truth is that most tech products never, ever live up to their potential. And Starlink, judged on its capabilities right now, is simply not a real competitor to the long, long coax wire running from my house to the local cable company fiber plant. It’s not even a great competitor to my data-capped-and-throttled “unlimited” AT&T 5G service because I can reasonably work from home on that connection and I really can’t with Starlink. And in the end, Starlink’s traffic has to run over fiber in the ground anyway.

It feels like we should all be more honest about what this thing can do, what we hope it can do in the future, and why our existing networks aren’t doing that already.

Let’s end this three ways. First, the team at Starlink should be legitimately proud of having shipped this ridiculously complex system, where they’ve gotten it to thus far, and where it might go.

Starlink is a truly remarkable feat of engineering, and the sheer force of will required to make it work as a simple consumer product shines through. It is, however, in everyone’s best interest to consider the trade-offs of having done all this work and putting all these satellites in orbit simply to get internet access. Astronomers and scientists are very mad about this. Starlink should talk to them more.

Second, all the people dreaming of Starlink upsetting cable monopolies and reinventing broadband need to seriously reset their expectations. At best, Starlink currently offers reasonably fast access with inconsistent connectivity, huge latency swings, and a significant uptick in time spent considering whether you can just get out the chainsaw and solve the tree problem yourself.

Maybe this will change as the company launches more satellites. Maybe it will eventually work better in areas that are dominated by tall trees. Maybe one day it will not drop out in wind and heavy rain. I didn’t give Starlink a formal review score because the whole thing is openly in beta and the company isn’t making many promises about reliability. But even when it’s final, you’re still looking at a service whose near-term, best-case scenario is being competitive with a solid LTE connection. I am no fan of cable companies and wireless carriers, but it’s simply true that my cable broadband and 5G service are both faster and more reliable than Starlink, and they will almost certainly remain that way.

And lastly, if you are a telecom executive or regulator in the United States, you have no choice but to see Starlink, its execution, and the unrestrained excitement and hype around it as a direct indictment of your rhetoric and efforts to properly connect this country to the internet over the past two decades. Dishy McFlatface is a sign that reads YOU FUCKED UP AND EVERYONE HATES YOU. Read the sign. This is your fault.

As a whole, the American telecom policy industrial complex has utterly failed to put fiber in the ground and signals in the air at fair prices and with good customer support. So much so that a total science project of an internet access system — which involves huge tradeoffs for scientific research and doesn’t work if there are trees in the way — has captured the attention and imagination of millions.

Broadband on the ground is so wrapped up in the lumbering bullshit of monopolistic regulatory capture that it seems easier and more effective to literally launch rockets and try building a network in the sky. Starlink isn’t the happy end result of a commitment to “facility-based competition.” It is thousands of middle fingers pointing at us from the air. It is what happens when there is an utter lack of competition.

Reality, as they say, bites.

Correction: An earlier version of this article referred to the Starlink dish as “Dishy McDishface.” The correct name is “Dishy McFlatface.” We regret the error.

Photography by Nilay Patel / The Verge