~7m37:10
Linus Tech Tips

Tesla “Hater” Uses Full Self-Driving for a Month

Nov 16, 2026

Read: ~7m · You save: 30 min

Tesla "Hater" Uses Full Self-Driving for a Month

A skeptic spends a month daily driving a Tesla with Full Self-Driving, logging wins, near-misses, and whether the subscription is worth $100 a month.

The author drove a Tesla equipped with Full Self-Driving (FSD) for one month, using a subscription-based FSD package. The vehicle belonged to Colton, who later drove it for a month as well. The author described FSD as Level 2 driver assistance that, under ideal conditions, performs more functions than a typical Level 2 system. The author recorded every drive with cameras and documented both positive and negative results.

Route Selection and Navigation

On the first day, the system delivered the author to a different house than the intended destination, with a blocked driveway. It also chose an alternative route home through a construction zone instead of the expected path. The author did not consider the mapping application at fault for the route choice.

The author later reported that the system handled routine trips with few problems. Two drives of more than 45 minutes each, to and from Richmond, ran without interruption. On another occasion, the system selected a destination the author had not entered. The author questioned whether the address had been set correctly that morning.

Merging, Intersections, and Traffic Flow

The system completed several complex maneuvers the author judged as correct:

  • Construction zone merge: The author described the merge as flawless. The system displayed temporary road markings in real time.
  • Four-way stop: The system stopped, yielded, and proceeded in the correct order when other rules were in effect.
  • Bus stop: The system braked when a bus signaled to leave a stop and gave the bus the required space.
  • Yellow light: The system slowed without a full stop, as the author would have done.
  • Pedestrian: The system stopped when a pedestrian stepped into the street.

The system also drove in the left lane and followed the vehicle ahead at a close distance during merges. The author noted that it matched the speed of surrounding traffic. At one point, the car traveled at 112 in a 50 zone, which the author attributed to keeping pace with the flow of traffic.

The author noted that the system did not enforce school zone speed limits, which the author considered a serious shortcoming. The author also noted that the system's speed behavior could create a hazard when it exceeded the limit to match faster traffic.

Parking

Parking proved difficult. In one lot, the system circled for an extended period and passed open spaces without parking. The author noted that human drivers also struggle in confusing lots. On another occasion, the system parked in a stall that the author judged as correct, after several attempts.

The system was also observed to misjudge parking spaces at a neighbor's driveway, where it repeatedly stopped and left the author at the curb.

Interventions and Safety Events

The author recorded several events that required intervention or raised safety concerns:

EventResult
Driveway misidentified on day oneDelivered to the wrong house
Parking lot maneuverAttempted to drive into an adjacent vehicle; the author intervened
Rolling-block eventVehicle did not slow for a bicycle and car, and the author briefly took control
Cut-in at an intersectionVehicle entered an intersection without a turning lane; a driver behind closed the gap, and the author was stuck in the intersection
Squirrel in roadVehicle detected it as an animal and slowed, but did not stop; the author applied the brake
Cyclist and scooterThe author reported that the system did not detect a rider in dark clothing without reflective gear, though it stopped at a green light for an unclear reason

The author stated that the intersection cut-in was the worst intervention of the month. The author also stated that the system responded correctly to a person whose path would have placed them in danger, slowing without hard braking.

The author reported that a camera-only system refused to engage on one occasion because the sun was blinding multiple cameras. The author cleaned the cameras, and the system then resumed operation.

The author also reported a software update after the squirrel event. The author did not state whether the update resolved the animal detection behavior.

The author estimated that there were about two actual dangerous interventions over the month. The author also stated that the system had potentially prevented one collision, when it stopped for a scooter rider the author had not seen. The author counted fewer critical interventions than the fingers on one hand.

Driver Monitoring

The system monitored driver attention through eye direction and hand position. As long as the driver looked at the road for most of the time, the system did not issue frequent warnings. The author said this made it easy to disengage from driving, which the author considered a safety concern. The author also reported that the system can turn off hands-free mode and require the driver to hold the wheel if the driver's hands are not visible, and that simply sitting in the driver's position caused hands-free mode to return within about 10 seconds.

Debris and Weather

Road debris had been a significant issue in earlier versions of the software. During this month, the system encountered debris only once, when an empty cup was on the road. It also encountered a low-profile debris object in the center of a lane and did not hesitate to drive over it.

The author drove in heavy rain on several days without the system disengaging. The author noted that a warning appeared during heavy rain or fog, but the system continued operating.

The author also stated that the rear camera area requires regular cleaning because the design does not seal it. The author considered this a trade-off, since it also allows the owner to clean the area quickly.

Software Updates and Features

The author reported several recent improvements:

  • A software update that the author described as making the system better.
  • Patch notes indicating work on a parking feature that adjusts to the owner's preferences, such as reversing into a driveway for easier charging.
  • A learned shortcut. After the author intervened several times on a preferred route, the navigation began using it automatically.

Cost and Value

The author stated that the system was impressive but difficult to value at its current price. The author noted that the system requires a high level of alertness, which the author considered difficult to justify at over $1,000 per year. The author also cited a subscription price of CAD 100 per month.

The author described the value of the system as tied to whether the driver could legally do other tasks while the vehicle drives. The author said that if the system allowed the driver to use commute time for other activities, the price would be easier to justify. The author estimated that a fair price would be $50 or less per month.

The author also observed that the navigation feature, which shows route and travel time, would be a reason to upgrade to a premium tier costing about $15 if it were restricted behind a paywall.

Conclusion

The author found the system significantly improved over earlier versions and capable of complex urban driving. The author identified remaining limitations, including school zone speed enforcement, parking in some lots, detection of low-visibility road users, and sensitivity to sunlight. The author stated that the system was close to operating autonomously in a way the author could trust, but that the current level of required attention and the subscription cost did not justify purchase for the author at that time.