DERSA X

Look inside

The decisions DERSA makes on your behalf, shown as the rules they are rather than described with adjectives. Move them and see what changes.

One. The fare

This one is not a demonstration. The prices below come from the same pricing engine the app books against, live, right now.

Inside town the fare is distance and time. Past 60 km DERSA refuses to price it as a town fare at all: that is not a town journey, and the driver has an empty road home. Ask for one anyway and you are quoted the intercity rate, because the engine decides what kind of journey it is rather than trusting whoever asked.

Two. How a courier is chosen

Everybody assumes closest wins. It usually does. The interesting part is when it does not.

Pickup

Courier A

9.84

From pickup0.25 kmdrag them on the map

  • Pickup ETA9.38
  • Capability fit0.20
  • Reliability0.00
  • Workload balance0.27
  • SLA risk0.00

Courier B

10.25

From pickup0.25 kmdrag them on the map

  • Pickup ETA9.38
  • Capability fit0.61
  • Reliability0.00
  • Workload balance0.27
  • SLA risk0.00

Courier B gets the offer. Distance usually decides, but not on its own.

The five factors are named. This decides who earns money, and “there is an OrderBy in a query” is not an answer to a courier who asks why they get nothing.

Reliability sits at zero because it is not measured yet. Zero means no evidence, which is a different thing from bad, and it will read as such the day the signal arrives.

What runs on this page is a faithful port of the rule, not the rule itself. The platform's own code is the authority.

Three. Why a shop is where it is

A closed shop is not a worse choice. It is not a choice.

  1. 1

    Kategna

    0.9 km · 18 min · 4.1 (120)

    0.689
  2. 2

    A brand new kitchen

    0.7 km · 20 min · 5.0 (1)

    0.677
  3. 3

    Yod Abyssinia

    2.2 km · 25 min · 4.4 (380)

    0.529

Yod Abyssinia

Kategna

A brand new kitchen

Close a shop and watch it drop below every open one, whatever its rating. Then set one shop to 5.0 with a single review: it does not jump to the top.

Being open is a gate, not a weight. A closed shop stays in the listing, because somebody planning for tomorrow still wants to see it, but it never appears above an open one however good it is.

And a rating is pulled toward the average until enough people have left one, so a single five-star review does not open the listing — while a genuinely new shop still gets a real chance to be found.

A paid promotion cannot buy a place in this list. Sponsored is a separate list and is labelled as one.

Four. The code

Four digits is one in ten thousand. That was never the protection.

The courier's phone

3/3

Type the code the person receiving reads out.

The customer's phone

The code is in the recipient's app. The courier never sees it — they are told it out loud, or they are not.

••••

In the product this is not stored anywhere. It is computed from the job, the stop, who is carrying it and when it was issued, so there is no column holding the answer.

Get it wrong three times. Four digits was never the protection; the attempt limit is.

Three wrong guesses and it locks, and the courier cannot wave it through. A new code has to be issued, and issuing one makes the old one wrong immediately.

It is stored nowhere. It is derived from the job, the stop, whoever is carrying it and the moment it was issued, keyed by a server-side secret — so there is no column holding the answer, and a copy of the database does not let anybody collect a stranger's parcel.

The code on this page is generated in your browser for the demonstration. The rules around it — four digits, three attempts, the lockout — are the real ones.

Why this page exists

Platforms describe these decisions with adjectives — smart matching, intelligent pricing — which tells a driver who got no work, or a shop that fell down a list, precisely nothing.

The rules have names and can be looked at. A model may one day replace the inside of one of these factors; what must not happen is discovering there was never a place to put one.