Manual for Road Editors

[MortenCopenhagen]

This is part 4 in a mini-series with tips on Editing Roads. Find the series index in reply # 1 below.

Based on the comments shared under part 3 I have tried to summarize and illustrate the tips to improve our road edit approval rate. This is the kind of guidance I wish Google Maps would provide to aspiring Local Guides wishing to venture into Road Editing. Thanks to @MarcoDavoli , @ErmesT , and @JustJake for providing highly valuable comments and suggestions.

The manual will need to be updated as changes in the road approval process occur. This version reflects the situation during the spring of 2024.

Overview

1. Do not delete roads

2. Do not fix the shapes

3. Focus on populated areas

4. Keep your road edits close together

5. Avoid obstacles

6. New roads must be connected to at least one existing road

7. Closing holes

8. Avoid Complex Road Edits

9. All roads in one edit must be connected

10. Bridges, Underpasses and Overpasses

11. Outlines can block new roads

12. Restricted areas

13. Too many issues in one area

14. Editor tips

15. Tips to spot missing roads

16. How to quickly open the Road editor

1. Do not delete roads

Removing roads is practically impossible to get approved.

A Googler once wrote “Roads should be fixed, not deleted” when explaining Road Mapper.

Removing this road was rejected repeatedly.

2. Do not fix the shapes

Refrain from trying to adjust roads = realign = reshaping.

I tried to get this road aligned properly. It got rejected.

Currently, Google Maps is focussing on getting missing roads added, so fixing existing roads has a lower priority. Hence such edits are less likely to be approved.

@JustJake reported having some successwith using the “Other” field under Fix a Road to get realignment issues fixed.

He added: “I’m not sure if keywords play a part, but I’m always clear about what thing the mapped road is violating, such as homes, properties, physics … It takes some time, but the fixers are very good at redrawing.”

@ErmesT mentioned that shaping can get approved if bundled with adding a new road. I have not been able to confirm this.

I spend a lot of time realigning all these roads in Venezuela. All to no avail. Too complex and only trying to realign a bunch of roads. I intend to try the “Other”-method in the area.

3. Focus on populated areas

Areas where a lot of people live have higher priority so your chance of approval is higher in densely populated areas.

You get a higher approval rate in cities rather than in the country side.

4. Keep your road edits close together

Google Maps sometimes gives priority to approving all edits in one area. So concentrating your road edits to a limited area might lead to a higher approval rate. At the same time this makes it easier to followup on your road edits to check if more needs to be done.

5. Avoid obstacles

A clear view of roads sometimes get obstructed by trees, buildings, shadows, bridges etc.

Here are a few examples: 1: Trees cover the view of a path. 2: The roof over a gas station is mistaken for a building, and black shadows make half of the parking lots invisible from the sky.

This path is not getting approved despite several attempts. This is proabably due to the bottom part being cast in shade.

Obstacles may result in the AI refusing your new roads.

6. New roads must be connected to at least one existing road

Adding roads in the middle of nowhere will not be approved. It makes sense from a navigation point of view. So you need to first connect a village to a main road before adding roads inside a village. @MarcoDavoli confirmed this: “A new road that ends up in “nothingness” is often not approved.”

Trying to connect new roads to a pending road is also fruitless. You need to wait for the pending road to be approved before you can connect other new roads to it. This makes road editing unnecessarily time consuming.

7. Closing holes

Missing road sections that connect 2 existing road segments are easily approved.

Here is an example of adding a missing road segment to connect two existing roads. This edit was approved in less than 5 hours.

@MarcoDavoli added: “I can also say that the creation of a new road joining two roads is almost always approved”.

8. Avoid Complex Road Edits

These are examples of complicated road edits that are very unlikely to get approved.

Notice the many road connections = intersections. This is likely to confuse the AI that screens our road edits. None of these road edits got approved.

@ErmesT shared this: Complex edits are a big no. I usually break them down into two or three steps:

First, submit the main roads

Second, submit internal roads one by one. You can add more than one road if the two roads are connected, but avoid adding more than two connected roads in a single submission. This way you can easily submit up to 5 roads per minute.

Example of a complex road system broken down into steps. Left: The unapproved complex network of roads. Then I suggested only one of the roads. This got approved. Currently, I’m waiting for approval of the the section shown in pic 3. The section in pic 4 just got approved. You can check the status here.

If you like points this strategy will earn you more points, but it also takes a lot more time and planning since you need to wait for the first roads to get approved before adding the next set of roads in one or two rounds. This requires some planning and strong commitment.

9. All roads in one edit must be connected

The first image will not be approved. You need to submit two separate edits as shown in pic 2 and 3. This makes little sense, but you make it easy for the AI to approve your work.

10. Bridges, Underpasses and Overpasses

If you need to draw a road without the new road connecting (snapping) to existing roads because they pass over or under, then you can delete the existing road temporarily as explained here.

11. Outlines can block new roads

If the standard view shows a building or airport runways where you need to draw a new road the new roads are unlikely to be approved. This is also the case where the building outline is wrong (like when a parking lot is mistaken for a building - see example 1 below).

Links to these examples Parking lot and Runways.

12. Restricted areas

Repeated and unsuccessful attempts to add obvious parking lots near airports hints that there might be some block put in place near airports. Check out these two examples: Kalmar Öland Airport and Kumasi International Airport.

13. Too many issues in one area

You may come across areas where existing roads are drawn completely wrong - not just off set or drawn with a sloppy hand. Mayby the operator was drunk or wanted to create a mess. Or the satellite images were properly aligned.

Cleaning this up can take months due to the delays in approvals and the difficulty in getting roads removed. Here is one such example:

The roundabout does not exist. And major roads have not been draw. Existing roads do not follow what is shown in the satellite image. In such cases it is okay to give up and just leave a comment like this one in the “Other” field:

14. Editor tips

Currently, the road editor does not allow us to change the type of road (local road, bike/walking path, etc). Nor is it possible to split a road into two separate road segments. Instead, use the “Other” option to get such issues reported.

Don’t mix Edit types and Road types in one edit.

While drawing roads you may need to move the map around. Try using the arrow keys for this.

Zooming in and out can be done with your mouse wheel.

Use these to undo and redo edits.

Is drawing roads neatly or roughly improving the approval rate? In my experience the approval system does not favor roads drawn with high precision (with many clicks) compared to roads drawn with fewer clicks. Below the Green road is drawn neatly while the blue road is draws with a lot fewer clicks.

There are two rare bugs you should know:

a. The satellite image in Maps can be old and different from Road Editor.

b. Not all approved roads show up in Road Editor.

c. Roads shown in the Road Editor are invisible in Maps standard view.

15. Tips to spot missing roads

Use desktop rather than mobile.

Scan areas in satellite mode at a zoom level of about 50-200 m.

Flip repeatedly between standard view mode and satellite mode to spot new roads as shown below.

Does the flipping shown above help you spot the missing roads in the parking lots? You can visit the place here to check if the missing roads have been added.

Try searching in Maps for “New Subdivision”, “Land allotment” to find areas currently being developed.

Searching at the edge of cities can also prove effective.

Another strategy is to zoom in using Maps view so the building outlines become visible. Just look for clusters of buildings with no roads like shown below:

Opening the Road Editor before spotting a missing road is another useful strategy since missing roads are easier to spot. This is because paths and parking roads are not always shown outside the road editor.

In many areas most roads have already been drawn, so adding missing parking lots and bike/walking paths is pretty much what is left to fix.

Here is one more trick:

While browsing the map in satellite mode you can see the small standard view icon in the bottom left corner. This icon is updated dynamically to show the left side of the bigger map. This allows us to spot missing roads more quickly and without changing the view.

16. How to quickly open the Road editor

1. Right-click on the map near the missing road. Then select Report a data problem

2. Click Add or fix a road.

3. Click Add a road or click on the road in need of fixing.

If you like keyboard shortcuts Ctrl+Shift+F will open the “Product Feedback” menu.

This concludes the draft Manual for Road Editors. You are invited to suggest improvements and additions in the comments below.

One final thought is one of learning by doing. The best way to better understand and predict the outcome of road edits is to systematically study the outcome of our road edits. How do we best please the AI so more and more of our road edits get approved? And we need to consistently apply such lessons learned in the future - and keep learning and sharing this knowledge. Unfortunately, the 75% rejection rate makes it very difficult to learn and improve as is the case with normal edits.

Cheers

Morten

Credits:

The cover image was created using DALL·E 2 AI with the prompt

“A van Gogh style painting of a quiet road in the countryside during summer”.

[MortenCopenhagen]

Tips for Road Editors mini-series Index:

Part 1: Creating Gmail Folders and Filters

Part 2: How to Keep Track of Your Road Edits

Part 3: Lessons Learned from Suggesting 72 Road Edits

Part 4: Manual for Road Editors

Part 5: Conclusions

Cheers

Morten

[MortenCopenhagen]

@ErmesT just started a new series with advanced tips and tricks for road editors. Here is an index:

Part 1: How to draw roads over or under another road

[SholaIB]

Let me quickly comment here @MortenCopenhagen

Thank you for this manual. Dos and don’t is what I as a roader go by as well. Thank you for highlighting them again. We will be kicking off in a few weeks and these will be of use to us. I really wish we have more tools to work with or even more considerations on road edits. For all the effort and time we put in. The returns aren’t great. Am starting to work at it again after a good long break. Lol

Are you planning to join our meet up on Saturday? Otherwise I would be happy to organise another one sometime soon. You could be our guest speaker as well.

Happy guiding

Cheers

[MortenCopenhagen]

Thanks.

Yes I intend to participate. But I don’t see the Meet-link yet. I guess we don’t need to sign up somewhere, Correct?

Cheers

Morten

[ErmesT]

Thanks for this great Manual, @MortenCopenhagen , that it is resuming in a single post years of experiences.

Here below my comments point by point.

A. Do not delete roads

Road Mapper, where your example comes from, is slightly different from road editor, although the basic principles are the same, and it is not always possible to add a new road without deleting another one, even if what I usually try to do is to just work to modify the existing shape before to add the new segment. Sometimes, even working in this way, Maps considers a road as “removed”. This is visible in the sequence of edits that appears before to click on the “Submit” button. Segments removed in coordination with newly added roads are usually accepted. I will share a screenshot when it will happens to me again.

B. Do not fix the shapes

I would like to add: not to make an edit “just” to fix a shape", but doing it in coordination with the addition of a new road is absolutely not a problem. Maps approves all shapes if modified in the way indicated above. But you have to be careful to one thing: I avoid adding, in subsequent edits, a road connected to a modified shape whose edit is pending. Any non-acceptance of the edit would in fact have a cascade effect in all the other edits too. Furthermore there is a strange situation in which a shape appears modified even if it is not, but this is a detail that we will eventually deal with separately

C. Focus on populated areas

This is one of the most important principle, because Google Maps gives priority based on the potential number of users of the new road

D. Keep your road edits close together

Yes. Maps is apparently approving edits in two ways: by priority and by concentration. Several edits in the same area can trigger the AI for a check of an area. On the point C you said “Focus on Populated areas”, and this is true, but what about an area that will be highly populated in the future? I added areas in remote locations simply by starting with dozens of edits, which in the days following the first approvals became hundreds, even if in previously deserted areas. Areas that are becoming new cities

E. Avoid obstacles

This is true, although in the past the AI’s ability to distinguish shadows was much improved. Now, however, the AI seems to have been given more power to decide, and roads with hidden sections are (again) often rejected

F. New roads must be connected to at least one existing road

This is a “must be”! see: 3000+, and counting. My tips for adding roads in Google Maps

G. Closing holes

Yes, two connection points are always better than a single one

H. Avoid Complex Road Edits

Keep it simple for the AI to approve it faster.

I agree that you need to follow the area and keep track ot the approvals to appends the new roads, but in term of final result what is more efficient? A complex edit pending for a month or two, and probably not approved, or a sequence of “quickly approved” edits?

From the already mentioned post

I. Try again

In fact this is what I always suggested but only for edits rejected after 90 days: Understanding rejected road modifications: how to turn a disappointment into an opportunity

J. Editor limitations

Normally in this case I report the road, and it works

K. Tips to spot missing roads

I will add some more tips during the next meet-up

M. Do not mix different types of road in a single edit

This is my additional tip. Adding a Local road and a parking lot or a pathway in the same edit is usually rejected. Better doing it in two steps

Hope this can help

Ermes

[JustJake]

This is a lot of info to digest, @MortenCopenhagen . Thank you for putting it together. Along with @ErmesT additions in the comments, I think you guys have set a great foundation for new Roaders to get going just from one post.

I’ll add that under ‘L,’ you can also hit CTRL+Shift+F to bring up the edit menu. Thanks to @Rednewt74 's tips on keystrokes (some Road Mapper, some Map Editing), I’ve also found that using the arrow keys to move the map around while drawing is extremely useful.

[MortenCopenhagen]

@JustJake

Thanks for suggesting Ctrl+Shift+F. Brilliant - one click shorter!

The Ctrl+Z and Y mentioned in the post do not work in the Road editor. So added an image with the icons for undoing and redoing.

I also cleaned up by merging related images into one.

Next, I will work to integrate the input from @ErmesT

Cheers

Morten

[TerryPG]

Thank you @MortenCopenhagen this was definitely the most informative post of the 4, and thank you @ErmesT for more great info. I am a newish roadie, only started in November but I am really enjoying it, gives me something more to think about than photos and videos, ( especially now). Thank you.

Edited to add this, I love the “open” roads edit shortcut ,it saves so much time, thankyou.

My little tip, I’m sure some of you may already use is Click on F11 button to show a full screen, I find it helps.

[MortenCopenhagen]

Hi @TerryPG @JustJake @ErmesT @SholaIB @MarcoDavoli

I’m happy to announce that I just finished editing the opening post in this thread to reflect more experiences.

One of the coming days I expect to also post a follow-up on the progress made by following the tips and tricks shared in this Manual. My hope was to improve the approval rate considerably from the 24% mentioned here.

Unfortunately, this will not be good news. Some days I have been able to push the approval rate up to 30%. This is not what I had expected and hoped for. So my intensive attempt to learn how to edit roads to achieve a much higher approval rate simply failed. I have stopped making a lot of road edits and will need to wait 7 days before the final approval rates per day will have materialized.

Find the post here.

Cheers

Morten

[ErmesT]

Tonight I will check yesterday’s data @MortenCopenhagen .

What you wrote in the “manual” is fully correct, but the approvals and the approval rate depends, how I said several times, of more factors combined together, and simplifying the results may not ber the best thing to do.

Here below the data related my my March 11th edits, analysed till March 16th

As you can see, only 53.15% of my submissions where checked in 6 days, while 46.85% are pending.

Considering only the data processed by Maps, my approval rate is 96.96%. Considering all the submissions, the actual approval rate is 51.53%

[MortenCopenhagen]

Thanks, @ErmesT

I’m thrilled to see your amazing approval data, I don’t doubt them, but I also know for certain that the tips discussed in this thread are not sufficient for me to reproduce your results.

I did focus on making follow-up edits for 2 or 3 days in parking lots where the perimeter got approved and only a ton of connecting roads needed to be drawn.

It is very surprising that you have that many pending edits as my observations consistently shows all road edits get resolved in 7 days.

All the best

Morten

[MarcoDavoli]

thanks for the update @ErmesT , @MortenCopenhagen

these are my last week figures…

Approval rating

Looking at the below pictures from the 11th of march I:

• my New Roads count increased by 1691 new roads
• I have received 2469 mail saying “Thanks you…”
• and 1578 mail of approval

So my approval rate is close to 64% and, at the same time, I lost 891 roads in just 8 days… Really sad. (In mid january I wrote that I lost in the past 4 months more than 5K roads)

Anyway, 64 is good but.

Approval timing

Looking at the next picture you’ll notice that the majority of the approval process has been done within few days.

and last

did you notice that now the approval mail are arriving during the whole day?

Take care

Marco

@TerryPG @JustJake @SholaIB

[ErmesT]

It is a long time, @MarcoDavoli , that I receive approval emails all along the day.

I think this is related to different teams at work that follow different procedures. In fact the “classic” emails that arrive at around 1:30 - 2:00 GMT contains mixed approvals while the others are more “aggregate” to areas.

I use a different method to calculate the daily ratio: Approved of the day / [Approved of the day + (Rejected of the day-Rejected for overtime)]

I do not consider the roads rejected by overtime as a mistake, because they were never analysed.

What I think, @MortenCopenhagen , is that we don’t have all the information that are used by the team to prioritize the evaluation process.

To be confirmed, a theorem must be replicable. The fact that it is not replicated may depend on many factors, and we certainly do not know them all. I will respond shortly on your post sharing my data for the last 8 days

[SholaIB]

Well @MortenCopenhagen

Thank you for these series of posts on roads. I have been very busy, but clearing up my table in préparation for full drivers on roads aa we begin TRAC.

In my own idea approval/rejections has always been up and down. It hasn’t ever had any fixed pattern. It goes up today and a few more weeks nose dive again.

Anyway could we maybe a few of us combined and compare data over the next few weeks on roads. Would that be of any help to the team? We could record the amount of roads added and approved maybe for 21 days @ErmesT @MarcoDavoli @TerryPG @JustJake @Lukas_1 @PrasadVR . Just a suggestion anyway.

Appreciate all your efforts to support TRAC.

Cheers

[MortenCopenhagen]

Hi @MarcoDavoli

You wrote:

I lost 891 roads in just 8 days… Really sad.

This is actually consistent with my observation that the timeout is now a week and no longer 90 days. @ErmesT just published data that suggets the period might be a bit more than a week, but nowhere close to the 90 days it was before.

It is of course not nice to lose our efforts but having a shorter turn around period also has a lot of benefits.

All the best

Morten

[SholaIB]

Is that a backlash to our request @MortenCopenhagen ? No it’s not nice to lose data and time !!!

Like a week is quite harsh or brusque. One hardly know what way to go with this !!! Anyway we will go ahead on TRAC. Too late to back out now .

Happy mapping guys

Cheers

[MortenCopenhagen]

Well, @SholaIB

I have just completed 3 weeks of intensive testing and documenting how road edits should be done to improve the approval rate. Sorry, but I think it is not worth it to keep trying if there is no realistic success in sight when it comes to improving the approval rate of road edits. I did my best, and it is not working. I tend to blame the approval system that now might rely on AI only.

I’m very open to admitting that I might have made mistakes or misunderstood the wonderful tips and tricks generously shared by our experienced colleagues. Thanks.

I can not help comparing road editing to suggesting “normal” edits on Google Maps. Finding missing roads is extremely hard. And learning from the feedback we get from the system is grossly hindered. It is not okay to have a setup where almost everything is based on guessing and the needed feedback takes forever to collect, analyze, understand, and use.

Google Maps will need to improve the approval system and provide the much-needed proper feedback and transparency to motivate me to put more effort into adding roads.

Sorry

Morten

[MortenCopenhagen]

@MarcoDavoli

You wrote:

Did you notice that now the approval mails are arriving during the whole day?

I have been keeping a close eye in this for 2-3 weeks now. And what I see it the following:

If I get 30 road edits approved per day only 2-4 of these come dripping at random times. The wast majority comes in batches within 15-30 minutes. This is often in the afternoon close to 2. 3 or 4 pm local time. It is not very predictable when the batch comes. But early or mid afternoon local time is very typical.

I tried to spot if the weekday vs weekend days make a difference - but I doubt it.

All the best

Morten

[SholaIB]

Well well @MortenCopenhagen

Well-done for the analysis done already. Sad that you found it not worth the stress. See you at the meet-up on Saturday right?!

Well it was worse than these I feel. I mean before the road editor was launched we never ever dreamed of 10k roads(talking to myself). @ErmesT is that true???

Maybe this will also improve sometime in the future.

Cheers