Community Diligence Review for Guazi Dynamic allocator

[Fatman13]

Type of Allocator

Manual

Allocator Pathway Name

Guazi Dynamic

Organization Name

Guazi Dynamic

Please provide the url to your Allocator Application

filecoin-project/notary-governance#1023

The Application number linked to your organization

https://github.com/filecoin-project/Allocator-Registry/blob/main/Allocators/1023.json

Account Authorization

Yes

How many Datacap refill requests have already been submitted for this pathway

0

1st Refresh - Please provide the url to your first application for DataCap refresh (submit for all that apply) (If this is the first, leave blank)

No response

1st Refresh - How much DataCap did you receive in this refresh application

None

1st Refresh - What compliance guidance were given to this pathway?

No response

2nd Refresh - Please provide the url to your past application for DataCap Refresh 2 (submit for all that apply)(If this is the first, leave blank)

No response

2nd Refresh - How much DataCap did you receive in this refresh application

None

2nd Refresh - What compliance guidance were given to this pathway?

No response

3rd Refresh - Please Please provide the url to your past application(s) for DataCap Refresh 3 (submit for all that apply)(If this is the first, leave blank)

No response

3rd Refresh - How much DataCap did you receive in this refresh application

None

3rd Refresh - What compliance guidance were given to this pathway?

No response

Please confirm that you understand that in addition to submitting this Github issue, you will be receiving a diligence review that will require you return to this issue to provide update clarification.

(Yes)

Please confirm that you have maintained the standards set forward in your application for each dispursment issued to clients and you understand the Fil+ guidelines and your role as an Allocator.

(Yes)

[filecoin-watchdog]

@Fatman13
Compliance Report

In the allocator’s application, the allocator said:

The overall diligence process will be automated with minimal manual interventions and falls in line with the ideology presented by FIP-0078 https://fips.filecoin.io/FIPS/fip-0078.html. We believe the proposed type of allocator will be first of its kind and will set a new standard for allocator models with exceptional features that address unmet needs in the ecosystem.

1. Could you explain how the automation was implemented? I don’t see what was automated in your applications at a glance.
2. You redirect users to FIP-0078, which doesn’t exist. What was the content of that file? Are you still using it to introduce new clients?
3. What was the KYB/KYC process? The issue indicates it was done, yet it's unclear how.

---

5 PiB granted to clients:

Client Name	DC
BOND	3PiB
TheSecond Life	2PiB

Example 1 - TheSecond Life #5

• The client didn’t confirm that the dataset is publicly available.
• According to the application, the first tranche should be 100TiB, while the first and only tranche was 2PiB. What was the reason for that?
• The client declared 7 copies, while there are 8 of them.

SPs provided:
f03079511 HK；
f02327534 US；
f02894855 CHINA；
f02826234 SG

SPs used for deals:
f02956383
f02826123
f02826234
f02948413
f02224274
f03079511
f01926635
f02901026

• Only 2 SPs match the provided list. The client didn’t update the SP list.
• 6 out of 8 SPs have retrieval at 0%, and the other two have retrieval below 40%.

Example 2 - BOND #8

• No additional questions on data preparation
• According to the application, tranches should be 100TiB, 500 TiB, 1 PiB, 1.5 PiB, while this client got 350 TiB, 768 TiB, 1 PiB, 921.6 TiB and 8.36 TiB. Why such schedule?
• This client has applied 12 times to different allocators with the same dataset. Sometimes at the same time.
• All SPs provided match SPs used for deals.
• Good retrieval rates.

[Fatman13]

Could you explain how the automation was implemented?

Most of the automation are done in the KYC / KYB processes and the rest are still under development due to lack of funding.

You redirect users to FIP-0078, which doesn’t exist. What was the content of that file?

FF seems to have moved the url which is out of my control.

Are you still using it to introduce new clients?

No. Clients expressed 0 interests.

What was the KYB/KYC process?

It is a combination of using alipay, wechat and qichacha to verify the legitimacy of individuals or enterprise.

The client didn’t confirm that the dataset is publicly available.

Those are mostly enterprise data confirmed in our email exchanges. Client wasn't familiar with application form.

According to the application, the first tranche should be 100TiB, while the first and only tranche was 2PiB. What was the reason for that?

That was my bad. Thats my first allocation and i typed in the wrong amount. The prompt gave the impression that its asking the whole amount while its actually asking for the 1st allocation amount.

The client declared 7 copies, while there are 8 of them.

Yes, that was communicated privately through chat messages.

Only 2 SPs match the provided list. The client didn’t update the SP list.

Client communicate with us through chat messages, they are not familiar with Github.

6 out of 8 SPs have retrieval at 0%, and the other two have retrieval below 40%.

Those are all private enterprise data which they don't expect retrievals.

According to the application, tranches should be 100TiB, 500 TiB, 1 PiB, 1.5 PiB, while this client got 350 TiB, 768 TiB, 1 PiB, 921.6 TiB and 8.36 TiB. Why such schedule?

Client has higher throughputs and seems did generally well after 1st batch.

[filecoin-watchdog]

The above explanations still leave some confusion on KYC/KYB issues. Do you have any reports/confirmations of these processes being carried out?

Those are mostly enterprise data confirmed in our email exchanges. Client wasn't familiar with application form.

That's right, it was an oversight on my part.

That was my bad. Thats my first allocation and i typed in the wrong amount.

If you have awarded too much DC, you can report this to the gov team and request to have your allocation reversed.

Client communicate with us through chat messages, they are not familiar with Github.

Do you have any records of these conversations?

Client has higher throughputs and seems did generally well after 1st batch.

But the first allocation was against your own rules defined in the allocator's application. Why were they formulated if you approach each client individually?
Whereas you clearly wrote then:
"29. Will you use standardized DataCap allocations to clients?:
Yes, standardized"

Those are all private enterprise data which they don't expect retrievals.

Also, why store data if you never expect to access it? This client said they want to access it yearly, which means the retrieval should still be good enough to provide it.

[Fatman13]

The above explanations still leave some confusion on KYC/KYB issues. Do you have any reports/confirmations of these processes being carried out?

I am not sure what you mean by reports? Every wechat and alipay accounts are backed up by a government issued id, by virtue of client able to delegate their account to us, it meant they own the account and by extension being the person behind the wechat account. KYB was done by clients supplying a series of government issued documents and compare them with informations on qichacha.

If you have awarded too much DC, you can report this to the gov team and request to have your allocation reversed.

I have created an issue right away at the time but was not picked up later.

Do you have any records of these conversations?

Yes

But the first allocation was against your own rules defined in the allocator's application. Why were they formulated if you approach each client individually?

i think at the time of allocator applications i would have no ideas of what clients actually wants or to better facilitate clients onboarding experience. My understanding is that as long as there rounds of allocation to prevent clients from abusing all DC, then i would put onboarding experience at higher priority.

"29. Will you use standardized DataCap allocations to clients?:
Yes, standardized"

My bad, i didn't get this part of the ldn template. Will remove this section in the future.

Also, why store data if you never expect to access it? This client said they want to access it yearly, which means the retrieval should still be good enough to provide it.

idk. My speculation is that since storage is free, the more copies the merrier? To clients, filecoin is just experiment and not mission critical. They could come back to check data integrity and see if they want to work with filecoin storage more. I will change the template to reflect such sentiment.