rfc9969.original.xml   rfc9969.xml 
<?xml version='1.0' encoding='utf-8'?> <?xml version='1.0' encoding='utf-8'?>
<!DOCTYPE rfc [ <!DOCTYPE rfc [
<!ENTITY nbsp "&#160;"> <!ENTITY nbsp "&#160;">
<!ENTITY zwsp "&#8203;"> <!ENTITY zwsp "&#8203;">
<!ENTITY nbhy "&#8209;"> <!ENTITY nbhy "&#8209;">
<!ENTITY wj "&#8288;"> <!ENTITY wj "&#8288;">
]> ]>
<?xml-stylesheet type="text/xsl" href="rfc2629.xslt" ?> <?xml-stylesheet type="text/xsl" href="rfc2629.xslt" ?>
<!-- generated by https://github.com/cabo/kramdown-rfc version 1.7.29 (Ruby 3.4. <!-- generated by https://github.com/cabo/kramdown-rfc version 1.7.35 (Ruby 2.5.
4) --> 9) -->
<?rfc compact="yes"?> <rfc xmlns:xi="http://www.w3.org/2001/XInclude" ipr="trust200902" docName="draft
<?rfc subcompact="yes"?> -iab-ai-control-report-02" category="info" consensus="true" submissionType="IAB"
<rfc xmlns:xi="http://www.w3.org/2001/XInclude" ipr="trust200902" docName="draft xml:lang="en" number="9969" tocInclude="true" sortRefs="true" symRefs="true" ve
-iab-ai-control-report-02" category="info" consensus="true" submissionType="IAB" rsion="3">
version="3"> <!-- xml2rfc v2v3 conversion 3.33.0 -->
<!-- xml2rfc v2v3 conversion 3.30.1 --> <link href="https://datatracker.ietf.org/doc/draft-iab-ai-control-report-02" r
el="prev"/>
<front> <front>
<title>IAB AI-CONTROL Workshop Report</title> <title abbrev="IAB AI-CONTROL Workshop Report">IAB AI-CONTROL Workshop Repor
<seriesInfo name="Internet-Draft" value="draft-iab-ai-control-report-02"/> t</title>
<seriesInfo name="RFC" value="9969"/>
<author initials="M." surname="Nottingham" fullname="Mark Nottingham"> <author initials="M." surname="Nottingham" fullname="Mark Nottingham">
<organization/> <organization/>
<address> <address>
<postal> <postal>
<postalLine>Melbourne</postalLine> <city>Melbourne</city>
<postalLine>Australia</postalLine> <country>Australia</country>
</postal> </postal>
<email>mnot@mnot.net</email> <email>mnot@mnot.net</email>
<uri>https://www.mnot.net/</uri> <uri>https://www.mnot.net/</uri>
</address> </address>
</author> </author>
<author initials="S." surname="Krishnan" fullname="Suresh Krishnan"> <author initials="S." surname="Krishnan" fullname="Suresh Krishnan">
<organization/> <organization/>
<address> <address>
<email>suresh.krishnan@gmail.com</email> <email>suresh.krishnan@gmail.com</email>
</address> </address>
</author> </author>
<date year="2025" month="September" day="06"/> <date year="2026" month="April"/>
<keyword>policy</keyword> <keyword>policy</keyword>
<keyword>Artificial Intelligence</keyword> <keyword>Artificial Intelligence</keyword>
<keyword>Robots Exclusion Protocol</keyword> <keyword>Robots Exclusion Protocol</keyword>
<keyword>web crawler</keyword> <keyword>web crawler</keyword>
<keyword>robots.txt</keyword> <keyword>robots.txt</keyword>
<abstract> <abstract>
<?line 87?> <?line 91?>
<!--[rfced] May we update the title to follow the format in other
workshop reports?
Original:
IAB AI-CONTROL Workshop Report
Perhaps:
Report from the IAB Workshop on AI-CONTROL
-->
<t>The AI-CONTROL Workshop was convened by the Internet Architecture Board (IAB) in September 2024. This report summarizes its significant points of discussion and identifies topics that may warrant further consideration and work.</t> <t>The AI-CONTROL Workshop was convened by the Internet Architecture Board (IAB) in September 2024. This report summarizes its significant points of discussion and identifies topics that may warrant further consideration and work.</t>
<t>Note that this document is a report on the proceedings of the workshop. The views and positions documented in this report are those of the workshop pa rticipants and do not necessarily reflect IAB views and positions.</t> <t>Note that this document is a report on the proceedings of the workshop. The views and positions documented in this report are those of the workshop pa rticipants and do not necessarily reflect IAB views and positions.</t>
</abstract> </abstract>
<note removeInRFC="true">
<name>Discussion Venues</name>
<t>Source for this draft and an issue tracker can be found at
<eref target="https://github.com/intarchboard/draft-iab-ai-control-report"/>
.</t>
</note>
</front> </front>
<middle> <middle>
<?line 93?> <?line 107?>
<section anchor="introduction"> <section anchor="introduction">
<name>Introduction</name> <name>Introduction</name>
<t>The Internet Architecture Board (IAB) holds occasional workshops design <!--[rfced] May we update the text as shown below (i.e., replace
ed to consider long-term issues and strategies for the Internet, and to suggest "large language models" with "Large Language Models (LLMs)", or would
future directions for the Internet architecture. This long-term planning functio this update change the intended meaning?
n of the IAB is complementary to the ongoing engineering efforts performed by wo
rking groups of the Internet Engineering Task Force (IETF).</t> Original:
<t>The Internet is one of the major sources of data used to train large la The Internet is one of the major sources of data used to train
nguage models (Large Language Models (LLMs), or more generally, "Artificial Inte large language models (Large Language Models (LLMs), or more
lligence (AI)"). Because this use was not envisioned by most publishers of infor generally, "Artificial Intelligence (AI)").
mation on the Internet, a means of expressing the owners' preferences regarding
AI crawling has emerged, sometimes backed by law (e.g., in the European Union's Perhaps:
AI Act <xref target="AI-ACT"/>).</t> The Internet is one of the major sources of data used to train
Large Language Models (LLMs) (or, more generally, Artificial
Intelligence (AI)).
-->
<t>The Internet Architecture Board (IAB) holds occasional workshops designed to
consider long-term issues and strategies for the Internet, and to suggest future
directions for the Internet architecture. This long-term planning function of t
he IAB is complementary to the ongoing engineering efforts performed by working
groups of the Internet Engineering Task Force (IETF).</t>
<t>The Internet is one of the major sources of data used to train large la
nguage models (Large Language Models (LLMs) or, more generally, Artificial Intel
ligence (AI)). Because this use was not envisioned by most publishers of informa
tion on the Internet, a means of expressing the owners' preferences regarding AI
crawling has emerged, sometimes backed by law (e.g., in the European Union's AI
Act <xref target="AI-ACT"/>).</t>
<t>The IAB convened the AI-CONTROL Workshop on 19-20 September 2024 to "ex plore practical opt-out mechanisms for AI and build an understanding of use case s, requirements, and other considerations in this space" <xref target="CFP"/>. I n particular, the emerging practice of using the Robots Exclusion Protocol <xref target="RFC9309"/> -- also known as "robots.txt" -- has not been coordinated be tween AI crawlers, resulting in considerable differences in how they treat it. F urthermore, robots.txt may or may not be a suitable way to control AI crawlers. However, discussion was not limited to consideration of robots.txt, and approach es other than opt-out were considered.</t> <t>The IAB convened the AI-CONTROL Workshop on 19-20 September 2024 to "ex plore practical opt-out mechanisms for AI and build an understanding of use case s, requirements, and other considerations in this space" <xref target="CFP"/>. I n particular, the emerging practice of using the Robots Exclusion Protocol <xref target="RFC9309"/> -- also known as "robots.txt" -- has not been coordinated be tween AI crawlers, resulting in considerable differences in how they treat it. F urthermore, robots.txt may or may not be a suitable way to control AI crawlers. However, discussion was not limited to consideration of robots.txt, and approach es other than opt-out were considered.</t>
<t>To ensure many viewpoints were represented, the program committee invit ed a broad selection of technical experts, AI vendors, content publishers, civil society advocates, and policymakers.</t> <t>To ensure many viewpoints were represented, the program committee invit ed a broad selection of technical experts, AI vendors, content publishers, civil society advocates, and policymakers.</t>
<section anchor="chatham-house-rule"> <section anchor="chatham-house-rule">
<name>Chatham House Rule</name> <name>Chatham House Rule</name>
<t>Participants agreed to conduct the workshop under the Chatham House R ule <xref target="CHATHAM-HOUSE"/>, so this report does not attribute statements to individuals or organizations without express permission. Most submissions to the workshop were public and thus attributable; they are used here to provide s ubstance and context.</t> <t>Participants agreed to conduct the workshop under the Chatham House R ule <xref target="CHATHAM-HOUSE"/>, so this report does not attribute statements to individuals or organizations without express permission. Most submissions to the workshop were public and thus attributable; they are used here to provide s ubstance and context.</t>
<t><xref target="attendees"/> lists the workshop participants, unless th ey requested that this information be withheld.</t> <t><xref target="attendees"/> lists the workshop participants, unless th ey requested that this information be withheld.</t>
</section> </section>
<section anchor="views-expressed-in-this-report"> <section anchor="views-expressed-in-this-report">
<name>Views Expressed in this Report</name> <name>Views Expressed in This Report</name>
<t>This document is a report on the proceedings of the workshop. The vie <t>This document is a report on the proceedings of the workshop. The vie
ws and positions documented in this report are expressed during the workshop by ws and positions documented in this report are expressed during the workshop by
participants and do not necessarily reflect IAB's views and positions.</t> participants and do not necessarily reflect the IAB's views and positions.</t>
<t>Furthermore, the content of the report comes from presentations given <t>Furthermore, the content of the report comes from presentations given
by workshop participants and notes taken during the discussions, without interp by workshop participants and notes taken during the discussions, without interp
retation or validation. Thus, the content of this report follows the flow and di retation or validation. Thus, the content of this report follows the flow and di
alogue of the workshop but does not attempt to capture a consensus.</t> alog of the workshop but does not attempt to capture a consensus.</t>
</section> </section>
</section> </section>
<section anchor="workshop-scope-and-discussion"> <section anchor="workshop-scope-and-discussion">
<name>Workshop Scope and Discussion</name> <name>Workshop Scope and Discussion</name>
<t>The workshop began by surveying the state of AI control.</t> <t>The workshop began by surveying the state of AI control.</t>
<t>Currently, Internet publishers express their preferences for how their content is treated for purposes of AI training using a variety of mechanisms, in cluding declarative ones, such as terms of service, embedded metadata, and robot s.txt <xref target="RFC9309"/>, and active ones, such as use of paywalls and sel ective blocking of crawlers (e.g., by IP address, User-Agent).</t> <t>Currently, Internet publishers express their preferences for how their content is treated for the purposes of AI training using a variety of mechanisms . These include declarative mechanisms, such as terms of service, embedded metad ata, and robots.txt <xref target="RFC9309"/>, as well as active mechanisms, such as use of paywalls and selective blocking of crawlers (e.g., by IP address or U ser-Agent).</t>
<t>There was disagreement about the implications of AI opt-out overall. Re search presented at the workshop <xref target="DECLINE"/> indicates that the use of such controls is becoming more prevalent, reducing the availability of data to AI (for purposes including training and inference-time usage). Some of the pa rticipants expressed concern about the implications of this -- although at least one AI vendor seemed less concerned by this, indicating that "there are plenty of tokens available" for training, even if many opt out. Others expressed a need to opt out of AI training because of how they perceive its effects on their con trol over content, seeing AI as usurping their relationships with customers and a potential threat to whole industries.</t> <t>There was disagreement about the implications of AI opt-out overall. Re search presented at the workshop <xref target="DECLINE"/> indicates that the use of such controls is becoming more prevalent, reducing the availability of data to AI (for purposes including training and inference-time usage). Some of the pa rticipants expressed concern about the implications of this -- although at least one AI vendor seemed less concerned by this, indicating that "there are plenty of tokens available" for training, even if many opt out. Others expressed a need to opt out of AI training because of how they perceive its effects on their con trol over content, seeing AI as usurping their relationships with customers and a potential threat to whole industries.</t>
<t>However, there was quick agreement that both viewpoints were harmed by the current state of AI opt-out -- a situation where "no one is better off" (in the words of one participant).</t> <t>However, there was quick agreement that both viewpoints were harmed by the current state of AI opt-out -- a situation where "no one is better off" (in the words of one participant).</t>
<t>Much of that dysfunction was attributed to the lack of coordination and <!--[rfced] In the last sentence below, please clarify what "both"
standards for AI opt-out. Currently, content publishers need to consult with ea refers to - is it new vendors and policy updates?
ch AI vendor to understand how to opt out of training their products, as there i
s significant variance in each vendor's behaviour. Furthermore, publishers need Current:
to continually monitor both for new vendors, and for changes to the policies of Much of that dysfunction was attributed to the lack of coordination
the vendors they are aware of.</t> and standards for AI opt-out. Currently, content publishers need to
consult with each AI vendor to understand how to opt out of training
their products, as there is significant variance in each vendor's
behavior. Furthermore, publishers need to continually monitor both for
new vendors and changes to the policies of the vendors they are
aware of.
Perhaps:
... Furthermore, publishers need to continually monitor both new
vendors and policy updates from the vendors they are aware
of.
-->
<t>Much of that dysfunction was attributed to the lack of coordination and stand
ards for AI opt-out. Currently, content publishers need to consult with each AI
vendor to understand how to opt out of training their products, as there is sign
ificant variance in each vendor's behavior. Furthermore, publishers need to cont
inually monitor for both new vendors and changes to the policies of the vendors
they are aware of.</t>
<t>Underlying those immediate issues, however, are significant constraints that could be attributed to uncertainties in the legal context, the nature of A I, and the implications of needing to opt out of crawling for it.</t> <t>Underlying those immediate issues, however, are significant constraints that could be attributed to uncertainties in the legal context, the nature of A I, and the implications of needing to opt out of crawling for it.</t>
<section anchor="crawl-time-vs-inference-time"> <section anchor="crawl-time-vs-inference-time">
<name>Crawl Time vs. Inference Time</name> <name>Crawl Time vs. Inference Time</name>
<t>Perhaps most significant is the "crawl time vs. inference time" probl em. Statements of preference are apparent at crawl time, bound to content either by location (e.g., robots.txt) or embedded inside the content itself as metadat a. However, the target of those directives is often disassociated from the crawl er, either because the crawl data is not only used for training AI models, or be cause the preferences could be applicable at inference time.</t> <t>Perhaps most significant is the "crawl time vs. inference time" probl em. Statements of preference are apparent at crawl time, bound to content either by location (e.g., robots.txt) or embedded inside the content itself as metadat a. However, the target of those directives is often disassociated from the crawl er, either because the crawl data is not only used for training AI models or bec ause the preferences could be applicable at inference time.</t>
<section anchor="multiple-uses-for-crawl-data"> <section anchor="multiple-uses-for-crawl-data">
<name>Multiple Uses for Crawl Data</name> <name>Multiple Uses for Crawl Data</name>
<t>A crawl's data might have multiple uses because the vendor also has <t>A crawl's data might have multiple uses because the vendor also has
another product that uses it (e.g., a search engine), or because the crawl is p another product that uses it (e.g., a search engine) or because the crawl is pe
erformed by a party other than the AI vendor. Both are very common patterns: ope rformed by a party other than the AI vendor. Both are very common patterns: Oper
rators of many Internet search engines also train AI models, and many AI models ators of many Internet search engines also train AI models, and many AI models u
use third-party crawl data. In either case, conflating different uses can change se third-party crawl data. In either case, conflating different uses can change
the incentives for publishers to cooperate with the crawler.</t> the incentives for publishers to cooperate with the crawler.</t>
<t>Well-established uses of crawling, such as Internet search, were se <t>Well-established uses of crawling, such as Internet searches, were
en by participants as at least partially aligned with the interests of publisher seen by participants as at least partially aligned with the interests of publish
s: they allow their sites to be crawled, and in return, they receive higher traf ers: They allow their sites to be crawled, and in return, they receive higher tr
fic and attention due to being in the search index. However, several participant affic and attention due to being in the search index. However, several participa
s pointed out that this symbiotic relationship does not exist for AI training us nts pointed out that this symbiotic relationship does not exist for AI training
es -- with some viewing AI as hostile to publishers, because it has the capacity uses -- with some viewing AI as hostile to publishers because it has the capacit
to take traffic away from their sites.</t> y to take traffic away from their sites.</t>
<t>Therefore, when a crawler has multiple uses that include AI, partic ipants observed that "collateral damage" was likely for non-AI uses, especially when publishers take more active control measures, such as blocking or paywalls, to protect their interests.</t> <t>Therefore, when a crawler has multiple uses that include AI, partic ipants observed that "collateral damage" was likely for non-AI uses, especially when publishers take more active control measures, such as blocking or paywalls, to protect their interests.</t>
<t>Several participants expressed concerns about this phenomenon's eff ects on the ecosystem, effectively "locking down the Web" with one opining that there were implications for freedom of expression overall.</t> <t>Several participants expressed concerns about this phenomenon's eff ects on the ecosystem, effectively "locking down the Web" with one opining that there were implications for freedom of expression overall.</t>
</section> </section>
<section anchor="application-of-preferences"> <section anchor="application-of-preferences">
<name>Application of Preferences</name> <name>Application of Preferences</name>
<t>When data is used to train an LLM, the resulting model does not hav <t>When data is used to train an LLM, the resulting model does not hav
e the ability to only selectively use a portion of it when performing a task, be e the ability to only selectively use a portion of it when performing a task bec
cause inference uses the whole model, and it is not possible to identify specifi ause inference uses the whole model, and it is not possible to identify specific
c input data for its use in doing so.</t> input data for its use in doing so.</t>
<t>This means that while publishers' preferences may be available when <t>This means that while publishers' preferences may be available when
content is crawled, they generally are not when inference takes place. Those pr content is crawled, they generally are not when inference takes place. Those pr
eferences that are stated in reference to use by AI -- for example, "no military eferences that are stated in reference to use by AI -- for example, "no military
uses" or "non-commercial only" cannot be applied by a general-purpose "foundati uses" or "non-commercial only" -- cannot be applied by a general-purpose "found
on" model.</t> ation" model.</t>
<t>This leaves a few unappealing choices to AI vendors that wish to co <t>This leaves a few unappealing choices to AI vendors that wish to co
mply with those preferences. They can simply omit such data from foundation mode mply with those preferences. They can simply omit such data from foundation mode
ls, thereby reducing their viability. Or, they can create a separate model for e ls, thereby reducing their viability. Or they can create a separate model for ea
ach permutation of preferences -- with a likely proliferation of models as the s ch permutation of preferences -- with a likely proliferation of models as the se
et of permutations expands.</t> t of permutations expands.</t>
<t>Compounding this issue was the observation that preferences change over time, whereas LLMs are created over long time frames and cannot easily be u pdated to reflect those changes. Of particular concern to some was how this make s an opt-out regime "stickier" because content that has no associated preference (such as that which predates the authors' knowledge of LLMs) is allowed to be u sed for these unforeseen purposes.</t> <t>Compounding this issue was the observation that preferences change over time, whereas LLMs are created over long time frames and cannot easily be u pdated to reflect those changes. Of particular concern to some was how this make s an opt-out regime "stickier" because content that has no associated preference (such as that which predates the authors' knowledge of LLMs) is allowed to be u sed for these unforeseen purposes.</t>
</section> </section>
</section> </section>
<section anchor="trust"> <section anchor="trust">
<name>Trust</name> <name>Trust</name>
<t>This disconnection between the statement of preferences and its appli <!--[rfced] May we update "was felt by participants to contribute to"
cation was felt by participants to contribute to a lack of trust in the ecosyste as shown below for easier readability?
m, along with the typical lack of attribution for data sources in LLMs, lack of
an incentive for publishers to contribute data, and finally (and most noted) a l Original:
ack of any means of monitoring compliance with preferences.</t> This disconnection between the statement of preferences and its
application was felt by participants to contribute to a lack of
trust in the ecosystem, along with the typical lack of attribution
for data sources in LLMs, lack of an incentive for publishers to
contribute data, and finally (and most noted) a lack of any means
of monitoring compliance with preferences.
Perhaps:
Participants felt that the disconnection between the statement of
preferences and its application contributes to a lack of trust in
the ecosystem, along with the typical lack of attribution for data
sources in LLMs, a lack of an incentive for publishers to
contribute data, and finally (and most noted) a lack of any means
of monitoring compliance with preferences.
-->
<t>This disconnection between the statement of preferences and its application w
as felt by participants to contribute to a lack of trust in the ecosystem, along
with the typical lack of attribution for data sources in LLMs, lack of an incen
tive for publishers to contribute data, and finally (and most noted) lack of any
means of monitoring compliance with preferences.</t>
<t>This lack of trust led some participants to question whether communic ating preferences is sufficient in all cases without an accompanying way to enfo rce them, or even to audit adherence to them. Some participants also indicated t hat a lack of trust was the primary cause of the increasingly prevalent blocking of AI crawler IP addresses, among other measures.</t> <t>This lack of trust led some participants to question whether communic ating preferences is sufficient in all cases without an accompanying way to enfo rce them, or even to audit adherence to them. Some participants also indicated t hat a lack of trust was the primary cause of the increasingly prevalent blocking of AI crawler IP addresses, among other measures.</t>
</section> </section>
<section anchor="attachment"> <section anchor="attachment">
<name>Attachment</name> <name>Attachment</name>
<t>One of the primary focuses of the workshop was on <em>attachment</em> -- how preferences are associated with content on the Internet. A range of mech anisms was discussed.</t> <t>One of the primary focuses of the workshop was on <em>attachment</em> , i.e., how preferences are associated with content on the Internet. A range of mechanisms was discussed.</t>
<section anchor="robotstxt-and-similar"> <section anchor="robotstxt-and-similar">
<name>robots.txt (and similar)</name> <name>robots.txt (and Similar)</name>
<t>The Robots Exclusion Protocol <xref target="RFC9309"/> is widely re <t>The Robots Exclusion Protocol <xref target="RFC9309"/> is widely re
cognised by AI vendors as an attachment mechanism for preferences. Several defic cognized by AI vendors as an attachment mechanism for preferences. Several defic
iencies were discussed.</t> iencies were discussed.</t>
<t>First, it does not scale to offer granular control over large sites where authors might want to express different policies for a range of content ( for example, YouTube).</t> <t>First, it does not scale to offer granular control over large sites where authors might want to express different policies for a range of content ( for example, YouTube).</t>
<t>Robots.txt is also typically under the control of the site administ <t>robots.txt is also typically under the control of the site administ
rator. If a site has content from many creators (as is often the case for social rator. If a site has content from many creators (as is often the case for social
media and similar platforms), the administrator may not allow them to express t media and similar platforms), the administrator may not allow them to express t
heir preferences fully, or at all.</t> heir preferences fully, or at all.</t>
<t>If content is copied or moved to a different site, the preferences <t>If content is copied or moved to a different site, the preferences
at the new site need to be explicitly transferred, because robots.txt is a separ at the new site need to be explicitly transferred because robots.txt is a separa
ate resource.</t> te resource.</t>
<t>These deficiencies led many participants to feel that robots.txt ca <t>These deficiencies led many participants to feel that robots.txt ca
nnot be the only solution to opt-out: rather, it should be part of a larger syst nnot be the only solution to opt-out: Rather, it should be part of a larger syst
em that addresses its shortcomings.</t> em that addresses its shortcomings.</t>
<t>Participants noted that other, similar attachment mechanisms have b <t>Participants noted that other similar attachment mechanisms have be
een proposed. However, none appear to have gained as much attention or implement en proposed. However, none appear to have gained as much attention or implementa
ation (both by AI vendors and content owners) as robots.txt.</t> tion (both by AI vendors and content owners) as robots.txt.</t>
</section> </section>
<section anchor="embedding"> <section anchor="embedding">
<name>Embedding</name> <name>Embedding</name>
<t>Another mechanism for associating preferences with content is to em <t>Another mechanism for associating preferences with content is to em
bed them into the content itself. Many formats used on the Internet allow this; bed them into the content itself. Many formats used on the Internet allow this;
for example, HTML has the <tt>&lt;meta&gt;</tt> tag, images have XMP and similar for example, HTML has the <tt>&lt;meta&gt;</tt> tag, images have Extensible Meta
metadata sections, and XML and JSON have rich potential for extensions to carry data Platform (XMP) and similar metadata sections, and XML and JSON have rich po
such data.</t> tential for extensions to carry such data.</t>
<t>Embedded preferences were seen to have the advantage of granularity <!--[rfced] Is "when it is moved" referring to "preferences"? If yes,
, and of "travelling with" content as it is produced, when it is moved from site may we update the text as follows?
to site, or when it is stored offline.</t>
<t>However, several participants pointed out that embedded preferences Original:
are easily stripped from most formats. This is a common practice for reducing t Embedded preferences were seen to have the advantage of granularity,
he size of a file (thereby improving performance when downloading it), and for a and of "travelling with" content as it is produced, when it is moved
ssuring privacy (since metadata often leaks information unintentionally).</t> from site to site, or when it is stored offline.
<t>Furthermore, some types of content are not suitable for embedding.
For example, it is not possible to embed preferences into purely textual content Perhaps:
, and Web pages with content from several producers (such as a social media or c Embedded preferences were seen to have the advantage of granularity,
omments feed) cannot easily reflect preferences for each one.</t> and of "traveling with" content as it is produced, when they are moved
<t>Participants noted that the means of embedding preferences in many from site to site or when they are stored offline.
formats would need to be determined by or coordinated with organisations outside -->
the IETF. For example, HTML and many image formats are maintained by external b
odies.</t> <t>Embedded preferences were seen to have the advantage of granularity, and of "
traveling with" content as it is produced, when it is moved from site to site or
when it is stored offline.</t>
<t>However, several participants pointed out that embedded preferences
are easily stripped from most formats. This is a common practice for reducing t
he size of a file (thereby improving performance when downloading it) and for as
suring privacy (since metadata often leaks information unintentionally).</t>
<t>Furthermore, some types of content are not suitable for embedding.
For example, it is not possible to embed preferences into purely textual content
, and web pages with content from several producers (such as a social media or c
omment feeds) cannot easily reflect preferences for each one.</t>
<t>Participants noted that the means of embedding preferences in many
formats would need to be determined by or coordinated with organizations outside
the IETF. For example, HTML and many image formats are maintained by external b
odies.</t>
</section> </section>
<section anchor="registries"> <section anchor="registries">
<name>Registries</name> <name>Registries</name>
<t>In some existing copyright management regimes, it is already common to have a registry of works that is consulted upon use. For example, this appro ach is often used for photographs, music, and video.</t> <t>In some existing copyright management regimes, it is already common to have a registry of works that is consulted upon use. For example, this appro ach is often used for photographs, music, and video.</t>
<t>Typically, registries use hashing mechanisms to create a "fingerpri nt" for the content that is robust to changes.</t> <t>Typically, registries use hashing mechanisms to create a "fingerpri nt" for the content that is robust to changes.</t>
<t>Using a registry decouples the content in question from its locatio <t>Using a registry decouples the content in question from its locatio
n, so that it can be found even if moved. It is also claimed to be robust agains n so that it can be found even if moved. It is also claimed to be robust against
t stripping of embedded metadata, which is a common practice to improve performa stripping of embedded metadata, which is a common practice to improve performan
nce and/or privacy.</t> ce and/or privacy.</t>
<t>However, several participants pointed out issues with deploying reg <t>However, several participants pointed out issues with deploying reg
istries at Internet scale. While they may be effective for (relatively) closed a istries at the scale of the Internet. While they may be effective for (relativel
nd well-known ecosystems such as commercial music publishing, applying them to a y) closed and well-known ecosystems, such as commercial music publishing, applyi
diverse and very large ecosystem like the Internet has proven problematic.</t> ng them to a diverse and very large ecosystem like the Internet has proven probl
ematic.</t>
</section> </section>
</section> </section>
<section anchor="vocabulary"> <section anchor="vocabulary">
<name>Vocabulary</name> <name>Vocabulary</name>
<t>Another major focus area for the workshop was on <em>vocabulary</em> -- the specific semantics of the opt-out signal. Several participants noted that there are already many proposals for vocabularies, as well as many conflicting vocabularies already in use. Several examples were discussed, including where ex isting terms were ambiguous, did not address common use cases, or were used in c onflicting ways by different actors.</t> <t>Another major focus area for the workshop was on <em>vocabulary</em> -- the specific semantics of the opt-out signal. Several participants noted that there are already many proposals for vocabularies, as well as many conflicting vocabularies already in use. Several examples were discussed, including where ex isting terms were ambiguous, did not address common use cases, or were used in c onflicting ways by different actors.</t>
<t>Although no conclusions regarding exact vocabulary were reached, it w as generally agreed that a complex vocabulary is unlikely to succeed.</t> <t>Although no conclusions regarding exact vocabulary were reached, it w as generally agreed that a complex vocabulary is unlikely to succeed.</t>
</section> </section>
</section> </section>
<section anchor="conclusions"> <section anchor="conclusions">
<name>Conclusions</name> <name>Conclusions</name>
<t>Participants generally agreed that on its current path, the ecosystem i s not sustainable. As one remarked, "robots.txt is broken and we broke it."</t> <t>Participants generally agreed that on its current path, the ecosystem i s not sustainable. As one remarked, "robots.txt is broken and we broke it".</t>
<t>Legal uncertainty, along with fundamental limitations of opt-out regime s pointed out above, limit the effectiveness of any technical solution, which wi ll be operating in a system unlike either robots.txt (where there is a symbiotic relationship between content owners and the crawlers) or copyright (where the d efault is effectively opt-in, not opt-out).</t> <t>Legal uncertainty, along with fundamental limitations of opt-out regime s pointed out above, limit the effectiveness of any technical solution, which wi ll be operating in a system unlike either robots.txt (where there is a symbiotic relationship between content owners and the crawlers) or copyright (where the d efault is effectively opt-in, not opt-out).</t>
<t>However, the workshop ended with general agreement that positive steps could be taken to improve the communication of preferences from content owners f or AI use cases. In discussion, it was evident that the discovery of preferences from multiple attachment mechanisms is necessary to meet the diverse needs of c ontent authors, and that therefore defining how they are combined is important.< /t> <t>However, the workshop ended with general agreement that positive steps could be taken to improve the communication of preferences from content owners f or AI use cases. In discussion, it was evident that the discovery of preferences from multiple attachment mechanisms is necessary to meet the diverse needs of c ontent authors and, therefore, that defining how they are combined is important. </t>
<t>We outline a proposed standard program below.</t> <t>We outline a proposed standard program below.</t>
<section anchor="potential-standards-work"> <section anchor="potential-standards-work">
<name>Potential Standards Work</name> <name>Potential Standards Work</name>
<t>The following items were felt to be good starting points for IETF wor <t>The following items were identified as good starting points for IETF
k:</t> work:</t>
<ul spacing="compact"> <ul spacing="normal">
<li> <li>
<t>Attachment to Web sites by location (in robots.txt or a similar m echanism)</t> <t>Attachment to websites by location (in robots.txt or a similar me chanism)</t>
</li> </li>
<li> <li>
<t>Attachment via embedding in IETF-controlled formats (e.g., HTTP h eaders)</t> <t>Attachment via embedding in IETF-controlled formats (e.g., HTTP h eaders)</t>
</li> </li>
<li> <li>
<t>Definition of a common core vocabulary</t> <t>Definition of a common core vocabulary</t>
</li> </li>
<li> <li>
<t>Definition of the overall regime; e.g., how to combine preference s discovered from multiple attachment mechanisms</t> <t>Definition of the overall regime, e.g., how to combine preference s discovered from multiple attachment mechanisms</t>
</li> </li>
</ul> </ul>
<t>It would be expected that the IETF would coordinate with other SDOs t o define embedding in other formats (e.g., HTML).</t> <t>It would be expected that the IETF would coordinate with other Standa rds Development Organizations (SDOs) to define embedding in other formats (e.g., HTML).</t>
<section anchor="out-of-initial-scope"> <section anchor="out-of-initial-scope">
<name>Out of Initial Scope</name> <name>Out of Initial Scope</name>
<t>It was broadly agreed that it would not be useful to work on the fo llowing items, at least to begin with:</t> <t>It was broadly agreed that it would not be useful to work on the fo llowing items, at least to begin with:</t>
<ul spacing="compact"> <ul spacing="normal">
<li> <li>
<t>Enforcement mechanisms for preferences</t> <t>Enforcement mechanisms for preferences</t>
</li> </li>
<li> <li>
<t>Registry-based solutions</t> <t>Registry-based solutions</t>
</li> </li>
<li> <li>
<t>Identifying or authenticating crawlers and/or content owners</t > <t>Identifying or authenticating crawlers and/or content owners</t >
</li> </li>
<li> <li>
<t>Audit or transparency mechanisms</t> <t>Audit or transparency mechanisms</t>
</li> </li>
</ul> </ul>
</section> </section>
</section> </section>
</section> </section>
<section anchor="iana-considerations">
<name>IANA Considerations</name>
<t>This document has no IANA actions.</t>
</section>
<section anchor="security-considerations"> <section anchor="security-considerations">
<name>Security Considerations</name> <name>Security Considerations</name>
<t>This document is a workshop report and does not impact the security of the Internet.</t> <t>This document is a workshop report and does not impact the security of the Internet.</t>
</section> </section>
</middle> </middle>
<back> <back>
<references anchor="sec-informative-references"> <references anchor="sec-informative-references">
<name>Informative References</name> <name>Informative References</name>
<reference anchor="CHATHAM-HOUSE" target="https://www.chathamhouse.org/abo ut-us/chatham-house-rule"> <reference anchor="CHATHAM-HOUSE" target="https://www.chathamhouse.org/abo ut-us/chatham-house-rule">
<front> <front>
<title>Chatham House Rule</title> <title>Chatham House Rule</title>
<author> <author>
<organization>Chatham House</organization> <organization>Chatham House</organization>
</author> </author>
<date>n.d.</date> <date/>
</front> </front>
</reference> </reference>
<reference anchor="CFP" target="https://datatracker.ietf.org/group/aicontr olws/about/"> <reference anchor="CFP" target="https://datatracker.ietf.org/group/aicontr olws/about/">
<front> <front>
<title>IAB Workshop on AI-CONTROL</title> <title>IAB Workshop on AI-CONTROL</title>
<author> <author>
<organization>Internet Architecture Board</organization> <organization>Internet Architecture Board</organization>
</author> </author>
<date>n.d.</date> <date/>
</front> </front>
</reference> </reference>
<reference anchor="PAPERS" target="https://datatracker.ietf.org/group/aico ntrolws/materials/"> <reference anchor="PAPERS" target="https://datatracker.ietf.org/group/aico ntrolws/materials/">
<front> <front>
<title>IAB Workshop on AI-CONTROL Materials</title> <title>IAB Workshop on AI-CONTROL Materials</title>
<author> <author>
<organization>Internet Architecture Board</organization> <organization>Internet Architecture Board</organization>
</author> </author>
<date>n.d.</date> <date/>
</front> </front>
</reference> </reference>
<reference anchor="AI-ACT" target="https://eur-lex.europa.eu/eli/reg/2024/ 1689/oj"> <reference anchor="AI-ACT" target="https://eur-lex.europa.eu/eli/reg/2024/ 1689/oj">
<front> <front>
<title>Regulation (eu) 2024/1689 of the European Parliament and of the Council</title> <title>Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU) 2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Artificial Intelligence Act) (Text with EEA releva nce)</title>
<author> <author>
<organization>European Parliament</organization> <organization>European Parliament</organization>
</author> </author>
<date year="2024" month="June" day="13"/> <date year="2024" month="June" day="13"/>
</front> </front>
</reference> </reference>
<reference anchor="DECLINE" target="https://www.ietf.org/slides/slides-aic ontrolws-consent-in-crisis-the-rapid-decline-of-the-ai-data-commons-00.pdf"> <reference anchor="DECLINE" target="https://www.ietf.org/slides/slides-aic ontrolws-consent-in-crisis-the-rapid-decline-of-the-ai-data-commons-00.pdf">
<front> <front>
<title>Consent in Crisis: The Rapid Decline of the AI Data Commons</ti tle> <title>Consent in Crisis: The Rapid Decline of the AI Data Commons</ti tle>
<author initials="S." surname="Longpre" fullname="Shayne Longpre"> <author initials="S." surname="Longpre" fullname="Shayne Longpre">
skipping to change at line 243 skipping to change at line 318
</author> </author>
<author initials="A." surname="Lee" fullname="Ariel Lee"> <author initials="A." surname="Lee" fullname="Ariel Lee">
<organization/> <organization/>
</author> </author>
<author initials="C." surname="Lund" fullname="Campbell Lund"> <author initials="C." surname="Lund" fullname="Campbell Lund">
<organization/> <organization/>
</author> </author>
<date year="2025"/> <date year="2025"/>
</front> </front>
</reference> </reference>
<reference anchor="RFC9309"> <xi:include href="https://bib.ietf.org/public/rfc/bibxml/reference.RFC.930
<front> 9.xml"/>
<title>Robots Exclusion Protocol</title>
<author fullname="M. Koster" initials="M." surname="Koster"/>
<author fullname="G. Illyes" initials="G." surname="Illyes"/>
<author fullname="H. Zeller" initials="H." surname="Zeller"/>
<author fullname="L. Sassman" initials="L." surname="Sassman"/>
<date month="September" year="2022"/>
<abstract>
<t>This document specifies and extends the "Robots Exclusion Protoco
l" method originally defined by Martijn Koster in 1994 for service owners to con
trol how content served by their services may be accessed, if at all, by automat
ic clients known as crawlers. Specifically, it adds definition language for the
protocol, instructions for handling errors, and instructions for caching.</t>
</abstract>
</front>
<seriesInfo name="RFC" value="9309"/>
<seriesInfo name="DOI" value="10.17487/RFC9309"/>
</reference>
</references> </references>
<?line 245?> <?line 330?>
<section anchor="about-the-workshop"> <section anchor="about-the-workshop">
<name>About the Workshop</name> <name>About the Workshop</name>
<t>The AI-CONTROL Workshop was held on 2024-09-19 and 2024-09-20 at Wilkin son Barker Knauer in Washington DC, USA.</t> <t>The AI-CONTROL Workshop was held on 2024-09-19 and 2024-09-20 at Wilkin son Barker Knauer in Washington, D.C., USA.</t>
<t>Workshop attendees were asked to submit position papers. These papers a re published on the IAB website <xref target="PAPERS"/>, unless the submitter re quested it be withheld.</t> <t>Workshop attendees were asked to submit position papers. These papers a re published on the IAB website <xref target="PAPERS"/>, unless the submitter re quested it be withheld.</t>
<t>The workshop was conducted under the Chatham House Rule <xref target="C HATHAM-HOUSE"/>, meaning that statements cannot be attributed to individuals or organizations without explicit authorization.</t> <t>The workshop was conducted under the Chatham House Rule <xref target="C HATHAM-HOUSE"/>, meaning that statements cannot be attributed to individuals or organizations without explicit authorization.</t>
<section anchor="agenda"> <section anchor="agenda">
<name>Agenda</name> <name>Agenda</name>
<t>This section outlines the broad areas of discussion on each day.</t> <t>This section outlines the broad areas of discussion on each day.</t>
<section anchor="thursday-2024-09-19"> <section anchor="thursday-2024-09-19">
<name>Thursday 2024-09-19</name> <name>Thursday, 2024-09-19</name>
<dl> <dl>
<dt>Setting the stage</dt> <dt>Setting the stage:</dt>
<dd> <dd>
<t>An overview of the current state of AI opt-out, its impact, and existing work in this space</t> <t>An overview of the current state of AI opt-out, its impact, and existing work in this space</t>
</dd> </dd>
<dt>Lightning talks</dt> <dt>Lightning talks:</dt>
<dd> <dd>
<t>A variety of perspectives from participants</t> <t>A variety of perspectives from participants</t>
</dd> </dd>
</dl> </dl>
</section> </section>
<section anchor="friday-2024-09-20"> <section anchor="friday-2024-09-20">
<name>Friday 2024-09-20</name> <name>Friday, 2024-09-20</name>
<dl> <dl>
<dt>Opt-Out Attachment: robots.txt and beyond</dt> <dt>Opt-Out Attachment: robots.txt and beyond:</dt>
<dd> <dd>
<t>Considerations in how preferences are attached to content on th e Internet</t> <t>Considerations in how preferences are attached to content on th e Internet</t>
</dd> </dd>
<dt>Vocabulary: what opt-out means</dt> <dt>Vocabulary: what opt-out means:</dt>
<dd> <dd>
<t>What information the opt-out signal needs to convey</t> <t>What information the opt-out signal needs to convey</t>
</dd> </dd>
<dt>Discussion and wrap-up</dt> <dt>Discussion and wrap-up:</dt>
<dd> <dd>
<t>Synthesis of the workshop's topics and how future work might un fold</t> <t>Synthesis of the workshop's topics and how future work might un fold</t>
</dd> </dd>
</dl> </dl>
</section> </section>
</section> </section>
<section anchor="attendees"> <section anchor="attendees">
<name>Attendees</name> <name>Attendees</name>
<t>Attendees of the workshop are listed with their primary affiliation. Attendees from the program committee (PC) and the Internet Architecture Board (I AB) are also marked.</t> <t>Attendees of the workshop are listed with their primary affiliation. Attendees from the program committee (PC) and the Internet Architecture Board (I AB) are also marked.</t>
<ul spacing="compact"> <ul spacing="normal">
<li> <li>
<t>Jari Arkko, Ericsson</t> <t><contact fullname="Jari Arkko"/>, Ericsson</t>
</li> </li>
<li> <li>
<t>Hirochika Asai, Preferred Networks</t> <t><contact fullname="Hirochika Asai"/>, Preferred Networks</t>
</li> </li>
<li> <li>
<t>Farzaneh Badiei, Digital Medusa (PC)</t> <t><contact fullname="Farzaneh Badiei"/>, Digital Medusa (PC)</t>
</li> </li>
<li> <li>
<t>Fabrice Canel, Microsoft (PC)</t> <t><contact fullname="Fabrice Canel"/>, Microsoft (PC)</t>
</li> </li>
<li> <li>
<t>Lena Cohen, EFF</t> <t><contact fullname="Lena Cohen"/>, EFF</t>
</li> </li>
<li> <li>
<t>Alissa Cooper, Knight-Georgetown Institute (PC, IAB)</t> <t><contact fullname="Alissa Cooper"/>, Knight-Georgetown Institute (PC, IAB)</t>
</li> </li>
<li> <li>
<t>Marwan Fayed, Cloudflare</t> <t><contact fullname="Marwan Fayed"/>, Cloudflare</t>
</li> </li>
<li> <li>
<t>Christopher Flammang, Elsevier</t> <t><contact fullname="Christopher Flammang"/>, Elsevier</t>
</li> </li>
<li> <li>
<t>Carl Gahnberg</t> <t><contact fullname="Carl Gahnberg"/></t>
</li> </li>
<li> <li>
<t>Max Gendler, The News Corporation</t> <t><contact fullname="Max Gendler"/>, The News Corporation</t>
</li> </li>
<li> <li>
<t>Ted Hardie</t> <t><contact fullname="Ted Hardie"/></t>
</li> </li>
<li> <li>
<t>Dominique Hazaël-Massieux, W3C</t> <t><contact fullname="Dominique Hazaël-Massieux"/>, W3C</t>
</li> </li>
<li> <li>
<t>Gary Ilyes, Google (PC)</t> <t><contact fullname="Gary Ilyes"/>, Google (PC)</t>
</li> </li>
<li> <li>
<t>Sarah Jennings, UK Department for Science, Innovation and Technol ogy</t> <t><contact fullname="Sarah Jennings"/>, UK Department for Science, Innovation and Technology</t>
</li> </li>
<li> <li>
<t>Paul Keller, Open Future</t> <t><contact fullname="Paul Keller"/>, Open Future</t>
</li> </li>
<li> <li>
<t>Elizabeth Kendall, Meta</t> <t><contact fullname="Elizabeth Kendall"/>, Meta</t>
</li> </li>
<li> <li>
<t>Suresh Krishnan, Cisco (PC, IAB)</t> <t><contact fullname="Suresh Krishnan"/>, Cisco (PC, IAB)</t>
</li> </li>
<li> <li>
<t>Mirja Kühlewind, Ericsson (PC, IAB)</t> <t><contact fullname="Mirja Kühlewind"/>, Ericsson (PC, IAB)</t>
</li> </li>
<li> <li>
<t>Greg Leppert, Berkman Klein Center</t> <t><contact fullname="Greg Leppert"/>, Berkman Klein Center</t>
</li> </li>
<li> <li>
<t>Greg Lindahl, Common Crawl Foundation</t> <t><contact fullname="Greg Lindahl"/>, Common Crawl Foundation</t>
</li> </li>
<li> <li>
<t>Mike Linksvayer, GitHub</t> <t><contact fullname="Mike Linksvayer"/>, GitHub</t>
</li> </li>
<li> <li>
<t>Fred von Lohmann, OpenAI</t> <t><contact fullname="Fred von Lohmann"/>, OpenAI</t>
</li> </li>
<li> <li>
<t>Shayne Longpre, Data Provenance Initiative</t> <t><contact fullname="Shayne Longpre"/>, Data Provenance Initiative< /t>
</li> </li>
<li> <li>
<t>Don Marti, Raptive</t> <t><contact fullname="Don Marti"/>, Raptive</t>
</li> </li>
<li> <li>
<t>Sarah McKenna, Alliance for Responsible Data Collection; Sequentu m</t> <t><contact fullname="Sarah McKenna"/>, Alliance for Responsible Dat a Collection; Sequentum</t>
</li> </li>
<li> <li>
<t>Eric Null, Center for Democracy and Technology</t> <t><contact fullname="Eric Null"/>, Center for Democracy and Technol ogy</t>
</li> </li>
<li> <li>
<t>Chris Needham, BBC</t> <t><contact fullname="Chris Needham"/>, BBC</t>
</li> </li>
<li> <li>
<t>Mark Nottingham, Cloudflare (PC)</t> <t><contact fullname="Mark Nottingham"/>, Cloudflare (PC)</t>
</li> </li>
<li> <li>
<t>Paul Ohm, Georgetown Law (PC)</t> <t><contact fullname="Paul Ohm"/>, Georgetown Law (PC)</t>
</li> </li>
<li> <li>
<t>Braxton Perkins, NBC Universal</t> <t><contact fullname="Braxton Perkins"/>, NBC Universal</t>
</li> </li>
<li> <li>
<t>Chris Petrillo, Wikimedia</t> <t><contact fullname="Chris Petrillo"/>, Wikimedia</t>
</li> </li>
<li> <li>
<t>Sebastian Posth, Liccium</t> <t><contact fullname="Sebastian Posth"/>, Liccium</t>
</li> </li>
<li> <li>
<t>Michael Prorock</t> <t><contact fullname="Michael Prorock"/></t>
</li> </li>
<li> <li>
<t>Matt Rogerson, Financial Times</t> <t><contact fullname="Matt Rogerson"/>, Financial Times</t>
</li> </li>
<li> <li>
<t>Peter Santhanam, IBM</t> <t><contact fullname="Peter Santhanam"/>, IBM</t>
</li> </li>
<li> <li>
<t>Jeffrey Sedlik, IPTC/PLUS</t> <t><contact fullname="Jeffrey Sedlik"/>, IPTC/PLUS</t>
</li> </li>
<li> <li>
<t>Rony Shalit, Alliance For Responsible Data Collection; Bright Dat a</t> <t><contact fullname="Rony Shalit"/>, Alliance For Responsible Data Collection; Bright Data</t>
</li> </li>
<li> <li>
<t>Ian Sohl, OpenAI</t> <t><contact fullname="Ian Sohl"/>, OpenAI</t>
</li> </li>
<li> <li>
<t>Martin Thomson, Mozilla</t> <t><contact fullname="Martin Thomson"/>, Mozilla</t>
</li> </li>
<li> <li>
<t>Thom Vaughan, Common Crawl Foundation (PC)</t> <t><contact fullname="Thom Vaughan"/>, Common Crawl Foundation (PC)< /t>
</li> </li>
<li> <li>
<t>Kat Walsh, Creative Commons</t> <t><contact fullname="Kat Walsh"/>, Creative Commons</t>
</li> </li>
<li> <li>
<t>James Whymark, Meta</t> <t><contact fullname="James Whymark"/>, Meta</t>
</li> </li>
</ul> </ul>
<t>The following participants requested that their identity and/or affil iation not be revealed:</t> <t>The following participants requested that their identity and/or affil iation not be revealed:</t>
<ul spacing="compact"> <ul spacing="normal">
<li> <li>
<t>A government official</t> <t>A government official</t>
</li> </li>
</ul> </ul>
</section> </section>
</section> </section>
<section numbered="false" anchor="iab-members-at-the-time-of-approval"> <section numbered="false" anchor="iab-members-at-the-time-of-approval">
<name>IAB Members at the Time of Approval</name> <name>IAB Members at the Time of Approval</name>
<t>Internet Architecture Board members at the time this document was appro ved for publication were:</t> <t>Internet Architecture Board members at the time this document was appro ved for publication were:</t>
<ul spacing="compact"> <ul spacing="normal">
<li> <li>
<t>Matthew Bocci</t> <t><contact fullname="Matthew Bocci"/></t>
</li> </li>
<li> <li>
<t>Roman Danyliw</t> <t><contact fullname="Roman Danyliw"/></t>
</li> </li>
<li> <li>
<t>Dhruv Dhody</t> <t><contact fullname="Dhruv Dhody"/></t>
</li> </li>
<li> <li>
<t>Jana Iyengar</t> <t><contact fullname="Jana Iyengar"/></t>
</li> </li>
<li> <li>
<t>Cullen Jennings</t> <t><contact fullname="Cullen Jennings"/></t>
</li> </li>
<li> <li>
<t>Suresh Krishnan</t> <t><contact fullname="Suresh Krishnan"/></t>
</li> </li>
<li> <li>
<t>Mirja Kühlewind</t> <t><contact fullname="Mirja Kühlewind"/></t>
</li> </li>
<li> <li>
<t>Warren Kumari</t> <t><contact fullname="Warren Kumari"/></t>
</li> </li>
<li> <li>
<t>Jason Livingood</t> <t><contact fullname="Jason Livingood"/></t>
</li> </li>
<li> <li>
<t>Mark Nottingham</t> <t><contact fullname="Mark Nottingham"/></t>
</li> </li>
<li> <li>
<t>Tommy Pauly</t> <t><contact fullname="Tommy Pauly"/></t>
</li> </li>
<li> <li>
<t>Alvaro Retana</t> <t><contact fullname="Alvaro Retana"/></t>
</li> </li>
<li> <li>
<t>Qin Wu</t> <t><contact fullname="Qin Wu"/></t>
</li> </li>
</ul> </ul>
</section> </section>
<section numbered="false" anchor="acknowledgements"> <section numbered="false" anchor="acknowledgements">
<name>Acknowledgements</name> <name>Acknowledgements</name>
<t>The Program Committee and the IAB would like to thank Wilkinson Barker Knauer for their generosity in hosting the workshop.</t> <t>The program committee and the IAB would like to thank Wilkinson Barker Knauer for their generosity in hosting the workshop.</t>
<t>We also thank our scribes for capturing notes that assisted in the prod uction of this report:</t> <t>We also thank our scribes for capturing notes that assisted in the prod uction of this report:</t>
<ul spacing="compact"> <ul spacing="normal">
<li> <li>
<t>Zander Arnao</t> <t><contact fullname="Zander Arnao"/></t>
</li> </li>
<li> <li>
<t>Andrea Dean</t> <t><contact fullname="Andrea Dean"/></t>
</li> </li>
<li> <li>
<t>Patrick Yurky</t> <t><contact fullname="Patrick Yurky"/></t>
</li> </li>
</ul> </ul>
</section> <!-- [rfced] FYI - We have added expansions for the following abbreviation
s
per Section 3.6 of RFC 7322 ("RFC Style Guide"). Please review each
expansion in the document carefully to ensure correctness.
Standards Development Organization (SDO)
Extensible Metadata Platform (XMP)
-->
<!-- [rfced] Please review the "Inclusive Language" portion of the online
Style Guide <https://www.rfc-editor.org/styleguide/part2/#inclusive_language>
and let us know if any changes are needed. Updates of this nature typically
result in more precise language, which is helpful for readers.
Note that our script did not flag any words in particular, but this should
still be reviewed as a best practice.
-->
</section>
</back> </back>
<!-- ##markdown-source: <!-- ##markdown-source:
H4sIAAAAAAAAA6VcXZfbxpF956/AoR+s2UNSirObsx7vR0YjyVI8Y81qxlGy H4sIAAHn82kAA81823IbSXbtO74ijX4Y0kGAurR7Wpw5M6YoqcUWKfGIlOU5
OXuSJtAkOwMCXDRAitbRL9q/sG/5Y3tvVXcD4IzsZOMHa0gAje76uHWruprz Dsc4UUgAOSxUwXUhhVboi/wL580/dtbaOzOrCqTU3aM+EZ6HaQpVlbVzX9e+
+XzSura059n0zcXz7OLN/PLt93fv3l5l7+vm3m/qXfbO7uqmnU5y09p13RzP ZE0mk1Hjm9wdmdPjp+b4dHLy5vXV2zdn5n1ZXdercmPeuk1ZNSM7m1Xu5mdv
M1et6smkqPPKbPFk0ZhVO3dmOTduntdV29TlvJGn5s++mvhuuXXeO1w47nA7 y2zjlmW1PTK+WJSj0bzMCrvG8vPKLpqJt7OJ9ZOsLJqqzCeVPDR58GhUtOuZ
3jOpuu3SNueTAmOeT/CMt5Xv/HnWNp2d7M+zX07u7fFQN8X5JJtnu7p0+ZF/ q47MkyffPRn5Df5qqrZuHj148ARX63a29nXt8dh2o7SO5njTkXn04NF3kwff
XTStW7ncmTJ7U7W2LN3aVrnlpXf1sm599vJDXnZ8W3bT1G2d1yUvHuwyyxtz jspZXeaucfXRqN3wEv7AW2pX1G0ti7nRxh+Zf2vK7MDU23XlFjX+wOv517+P
KG3Dj43cu2g/tJPJzuEdWZbX253J2/PsaD0/Y9qjryamazd1w/nwMoSA6V4v sLXHo9wWyyPjitG1296W1fxoZCZmU+Y+2/Kv46rxC595m5vTonF57peuyBwv
su/rtnXVemO28rWK5No096dXdrVvTSnvwn/z7NqWy7prKivf5HUHwUG4F51v vS1nZVOb5x+yvCWd5qIq8aoy58VbNzNZZW9zV/Gfldw7bT40o5Ftm1VZ8T0j
G1M6I1/brXHlebat6vbX/N+isq1c6Bp3nm3adufPnz49HA6LePXpaIK3i+y7 g//5AsSeT83rsml8sVzZtfy8aPNc2Xhuq+vdq5lvwO9zl8/Ktiqc/la2YDF+
xvlNZarB9G67xvrN+Ep4k5dLi/tw6ddrfr2AJCaTqm62pnV7aGxCE+g/Zdnl PgYXK5t7Kz+7tfX5kVkXZfPP/L9p4Rq50FZgzqppNvXR4eHt7e00Xj0ckHY5
64u71xfX89dvf7h9qWsMdnW5MS1EkL2uO2+zd12p621Ns7bteAm53rrhnYu6 Na8qX68KW+wQdtlWrl4Nr4a31XJpeh0u/fOSP0+zcj0aFWW1to2/cUejERWm
WT81EFA77/zTcGUul+ZNHCSpREUa/s0yPHvyYpnjq5vRzGjwycphL73xPzpD +5cxJy+Pr14en09evnl3+fxIFgyqerKyDTZvXpZt7czbNtc9N7Zauma4jUxv
2KqBZvJ72yycbVcyw3VTd7unxgWzP3id89Ofnx7NF8pvYdL5xrU2byH57Hlt XfHOaVktDy2Y1Eza+jBcmcilSRUXUbVa2LzWfycByf8m4b/GYK0dQoTmFxcD
moKTvbm4efnu9q+cL+wNg8Ep/N8z820c5O+fPaZ2cXk3mv07u+5KGAwm/sR2 SmktyUSgFZ3l3EsxXm4hrezaVVPvmoVQvKzKdnNofTCa21r3cPjryaXSQkEa
Z9lXz776x6e/+NU/f53Vq6zd2Oxl19Q7a+C3poH5b23VZqYq4uVLOEjuykcX KHK28o3LGkjGPC1tNSfxF8cXz99e/kL6oYlYDKZQf81O1nGR3343IPX45Gqw
aLtmXtoPC8shDP55akv3tLHrp+ktT+s///yqHpmC3CAwJTOeP/vV/Be/5BJf m7du2eZQMGxk7/m7fTqPbw8ffvf9E1MuTLNy5nlblRtnYb22gsmsXdEYW8zj
vLy8evP92Nanl4JkLXwvu4QHOXjgHeb+zuxckb2weekqGxd08SZ7AcVgYdst 5RMYVeZz/vPhY/NjWzhZwuR2C3M08/K2MCtbrcvC125uqFM1uWY7v+F7fkOW
Hpt+1i2SynzpCuvDP/OB6uaKoADhap7Le+d4w7zha+eFvnZer+RLYDTtAY/I 5kvmfLojrgZ1J/uwcvP4wYNDuMDvD5Re/PLwu9/jl4eP+7983/8Ff35/+P0/
a+fPni12xeonZROR5Kqu1rvGpu8DmGzMEasaXzx5+N0CJroxjTt5Fphtm3Z8 fd//58OHj5/Iy+JPTw4fPfz2W/npma/AOdhczQvfHj55cPj8Xffwd4e/f/L7
7eTRC7zXnr7zonG2HHx/8swlnumq4uShS7PdLREy+mtJr/80mczn88wsib05 /rOPHhw+/P7R92bvM87QHGfNvtm7ch8ac+ublXn+/NhULnc3Flf371UW11aT
gsKdKOhhMDwYD6iu9rayRbY8iiJ/wheyJ3DaM5rDrd21lkFPzGgBs3A+0yAJ 3H2YOrLf4j+HLveHlVseJgkdln/7eY24R3w9teJakwffTR4+pno8e35ydvp6
zN1usf4frc8cIph364pBzsCQdrWr8BVMpnA+7ySMilfABCqGQjzT1juX4x/g 6FfGJxIxGkjJnMBbeXi8K8j9rd14cMpluYfIgzIcn5pnUHIoxRoCr8efdUFJ
XbY1R8yxafjsqmswvYbz9bi9Uefjwwis94vJBAHK6mMtZ4OA3onX4W8T54Yn /evcz10d/jPpmcFEIxXCYzHJ5L0TvGFS8bWTub52Ui7kR0RP2hYekddOHjyY
uMZdU+fWFghnPtrvIchkkYmN7509eBkdgc7xVf2IkJWr9CVhWNPwzbW3p6Nl buaLL/Imeu6zslhuKpd+D457ZbfY1fDizsNvpzB3aLbfeRZRz1XN8NrOo8d4
O4MQn7ud4bo5XFFnCG1ZZXPrPcRUHjHIqoSYBREfee9Ctbl1RYFAMfmCGmrq r9t953HlXd77feeZEzzTFvOdh07sejODnnXXklz/aTSaTCbGzhjrMoTVP/7D
ost5VXX78zrb1GWBtea5oczBN+IMsSxLFWFRbZ2Em5Ww/zkG3UJ8vrM6JZoT ZPJv1SJz838HdVtEYKOwQEQn4jZNaRZlnpe38pvGHkq+xD+r0W30eopR6j+P
WBO1hHg5spaZ3IARfLdeW091yRwK19hcxXf6CKTWTzaYUP/aXWmqCgrCQJUM Rm8qv/SFzYXDP4ODRheuWlnoAO/V38yiKtfyri9EhcnkT6PRlajX3bVvbY3A
EEVLKTkvBKe0VIdpjnwxr+Hpms/Yag2MAPzz7xVeDOHvbMMgr4bO1fOihI1k Xty4At5lttW1Pu8FzR5etM8tXbpN4wiwxAimUGpfh40hOq/XkN5PcAAeCKb2
A2lmLwfP3xl/n72qm9xClC/vXp0tTmSOydQ9Dm7Nn7FODyoEBYulExURs0U6 y4J2bWEGmxLuqqbCz32dtQLAxBdAgQtaP55pyo3P8B9EQrMmp21V8dlFW5GP
ECBMpyQs4v/VujP4Y1sXtvTZkyv5+ip+fR2/vrr2ZzNAOm6ERMEQYfxleZxl pLfG7ZW6XT5Mzk5HI4AYp481pAZAsRV/i79tpA1PcI+bqsyco5eso/VF8UyN
08/wx+zJxZuz6dkie25zQ54itso/6PA0P1vtHQ1BpbEFl8t23bIES7KNzDox WOiNd7e1rL4pa69ONK4IXvlCXxKWtRXfXNZudzWzoavO/MZy31xuXhoAIVO4
Igq+OlV1tkVYkfvsB6Ak/BiCEgUcMDn/JRzMrmzDudBH1rBD3oEYIayVf28w zNU12JRvscgiB5tFive8d6q6uPbzOSDF6BtKqCrnbcarP6uZdJuQMohBLIG2
E6gPKy5mENfWtm6Lm5eM6zKp0hwQWhfrxUy9bRBSf6gwrS89x7uA33z8qAH6 Qzv3/NRND0h7bgE2xzl9jCFUbS3+WJdzl9djdbbjM7l4Fi+ey0Wzd3Z2Xu+P
06ekG9hIwrX2M+CHZf3i6/lXz07AjGqaYlklhb0jfgK7yqzetXNQIawc7K1y D+Amsdc2n4+EHeHFQELFUt/P8FTMwbE13Cj4vav1V32N8wxuiYdr+zesXgOC
fqsGjSnQ8JedKwv8lQGLIYAW33GREBClDqezfgZB/HcHd6DNevWX+iGi+YQt Zk51ht6xZSCEpcEuvYDCe6kHgV+gW8hel+qcEGSgTHm+PTDjz4WgvePT/fH+
HmzdTrE6ML1PnxaQf0ATEI9mJssSAfJFYZ5W3xi18dkcAoP++7tXl1//8tnX dGiEX0/6l2g0e2V1IET2KewRKD5jl0jQmAz+5w15VeZzkJdlloaINaPaQtcd
nz5lDBWlr7P7CurLoJdpn1BMeXUTrGZpbYXZ1tSmIQwubXvgd1GxWDqX6buS 7VbpjRZncrj0CRZdY7t161RP6SGRotF04fIGLuRAbsAKdbtcupo2LDTMFSLQ
qQJXkta2LAkGq2QVuLapD5wmvLexQG3XLrJXCvI09NkgrZFAQAfAPzoR2KDv pnYfgSl1xAa/0r0WCltQi7BQIQtEhtN0PH3ZepM72qittsJqXMPTJZ9xBdTO
XCvDHswxYBf5w3AyC9Dlg91biGsQcKITlG7r2jHumYg0/btVU2aHmGHyDX1a AR3y78WCPthsXEU/rd6Pu+dFQZXJMSTKnveev7L1tXlRVmT76fOrF9SNr9OH
lIZAUyWjOGBVaQxb0ArrjDlgQ0CojgLrIfjJvYgbEJPEklmMSOsG9J48xrWt v0ePzc+oyD3qYZ66zDLJUXvFH4wB9EiuuPFUA+XFuoS4Nu0sR4rlKqE5pVNk
tRDQXmZnsiXeDPC1pe2BEFZYiV3CUsE5IHYsG/Ze1FQBRcHQ1/s0vnN7V8LT e7EraDFwuc99QNiHawebhP23IK3+HXyuW7iKpNBtLqGFvAOgRxJZ/r0CJRAe
chCwY2aKfc0UOBiipqRbc0+hTRBkvngs05ncjKLZurFJegxH48gnnqBc98FI 9jtnIr12jV/j5hlBvxCVWzgvN13CefliiK/fFSDrdzXXA140Hz8qWv/0KUkG
NOphivXpEzFgFFKL2qqWTNs2btkhtsOtWvUevtTBxfau6GC6NA0wSrjlj8GH GpJCXfOZeIhtPXwyefRgJ75RSGNsKyerNwQECGdA7ZtmgrwJO6fD8/Va1Rkk
Dg7hGIoJ8ETwD1k7qBvRrk/jfQweaeaiIRFdrlFt0/k0DVrcN2q2jPoC7Bs+ UO1nrc+BxgsDcAEGNFZRORhErsPkXE0f/J8tjIEaW6u1lHeDXJ3CTb2Bwx5j
gFGgRMzIcnBCANyRj4syPrQQ68ePGAU6stbD7aAZruRzbGEGCZacu7yL2IGw d0gLP32agv8hwADoQxW4LWEgXxTodPrGKI3PlhWw6J/fvjh58vjBk0+fDLFP
KnAWGc4Qo+ERXPPGlrQ+qO+3QiRe6voHjEXrH4TJv4cj/f8okk2zKbomglRa XpfmumDYgFzGXY1hzKuroDUz5wpQW1KalpFx5ppb/hYFi61zm3Wbs8LAnaS9
PDD/b2RLCAGP86URjPAl0SHCQsKc4GykMk29zYI7BvNZO7hS5AmP8zhMi2QV zXK6gkXSClxbKXiC7VaO4KmZmhca96nmB71Kh2ADunH8RwmBDtatb2TZW7sN
HlMNV9PDDBQYrdAxcuINbYCXJtsbZD3yibLs/COz7EW3qsuyPqiprPCXSgZB nouAuE/MFLn1rbtxYFcPg0QjyP3aN0OvZ6Of6d6tkrIbwAibrWjRIjRgjyIp
v153DxknbHTkPHa7a8VJzU4YmclSgYqW0sfD2xzRVcZ+kRah4bQfG6Fc5AJk xS12ldZwc2phaVg8qugOiq1E+oCH5F6EY7BJ4MVBBCnLyq7p1EBU4xhMb4Q6
29tjXLQ4JidC6FUUxtCXXQOsb0lUElcakIzomhjANSPCwLgaYoNrklAgD4kT a2Z4M1yvy13nBqGFheglNBUgGmzHtqHv85IiICuIhjqbxm/+BslnXWbIKLbG
sB7esOsaaFwpFl4rxIrz0ehnIOFGgA6X+5BNKoFYKLGZiaRppOZD/oZrvss3 zm9KVtuCImqVam2vybQRcMc395VJRhcDgLOsXOIeEcoQDIklaOJ7ZyUqdb8+
DH/knzKut80eQXWWkR0UBd4MmmJI6BQ1BwHq48cUT0PAyB8ZudMMYWeOBzC4 8+kTfcAAZc1Lp1KyTVP5WQukAbNq1Hr4Ug8Tu/HzFqpL1UCKBLP8KdgQ4QwF
QKoV13Hvsqzz+0AbYgSLFAgyf3MD1C4os1n2A6Y2vwDZawPlaZTbwfwElrXc E9wTXX8oASIXobfraoJ1DB2JcpGQsC7TmLZq60QGNe4PqrYEguLWV3wAq0CI
wNqNKMiBKyNaqHGrwGLoqvfCJhcABm/JyLMUmTJzAukfP4YyAdCL8CsBJAKS oMhxcbqAkKOLMD40YOvHj1iFCMnVMDtIhjv5HIA8AAdz0i7vou9AUBV3FkFv
jauTxQYz8FTb0sLRuK6tUioL68cLyBQQNqIRmb1xpVm60qnWhDfDbjHXJyOF 30fDIrjnlcupfRDfvwi2fK77VxAr8TVmGFdfBZv/PtTsEjXztopOKm0ePv/X
9ypMSpcUsgoGNCeTxGTAoUGBb+Hj0UdGTtyDEeaawzx/Ql7ijUKT6NDrDSVT AOiABBAG7ofRA1fCm6NRhM0EumBwBDNMroJJBhVaephTRAr3w3uQxhwGVlP0
WuNbYf4p8kKXlgmGwHYYNSbWToxPZKYrxhDTVhRHaNxRIrLutgak+CiOEgxQ d9S5GggxaiKxcYU3NMHFVObGIpWXf5GfbX0PlR37NNdUdVkQ1wt3EPfL5Z00
EqewVNgiocmtlFdAixkmvcjetkOnEtpQhegc7jl1lGXIDfB1ImIIk7mlKTJx BFo6MB+33jRipnYjiMyaVNumrnQR8TJDfA11nbgFDajd2gjmwhX4thu3jVsW
R+IEw/QhHgRPJMGiyUS3nHHNgduLgUNLQaN4orFaI/Mbt9OgnAFYWqikUeM3 0yQhdL7qh7H0SVvB2zcEKgkr9WBGNE4s4KsBZGBkDdHBV4kl4IZECuhPRJKb
gG0Owwym3QgNxIwPyFZJgAoWj5FtwsQTj2uTrYNM5/dZb+4iUjjk5gHb2piY toLMFWbh1QKuSJPGQAseV+LucLkL3KK+NT0s4iLMlTUSW0npuHcXPFKbrRgO
9wnKKi6NUCt6A3WcIY50itQHedm0qkXPYsoAVJCN1WqaPQlpCev6YiW8Z2Bi iUblDbWrbhBkDwzRwlwTncYS3qkX7QWsjx9TfD3gGrcsPeC/Nvvsa1rNKDd2
9Mxr+oIYEOZWHH1KZrmAxGyKSEFKZD/i/ZFjxxKHZBSG7wlJR5jwIhug7EPW ewtwF/C2On08McvL7DpgihjeIj6COE4v4NLnwk6w5h0onRwDCjYBEVUK/aCZ
l2yAkA9Sriqw4LEDo8XlPmtRWxgZTbKYiNRSeyBx9EEbblzlIfQK+4GA5F36 4rW1NMk6sOZtANIIJqr3yskY2cobgZpT+I3aEa6bFLiM3fH4Hz+GshicG72z
oi8pvo3ZO+TGJ0z/8RnDUTomusCNyrWYqSiXAqjsoSe7nDW/JMivbaJzQmmd xJfor1zcn2w36EhNmc4cbJA7WyvicjAMvIBAAlElapi9sT63M597FaeAaig1
TeQl3N9TN3Pg/+sV1PQD11+GeMYyjtvCYBztQ8sfM8pFDZAPDVdL0YqI2gCF aN2jNiRNUKnKY1EbpOhQBO2aEGiCGABsIORLmH80oIF9d74KtGbQ3S/wSwxV
ed2VhaQoI/12hIOW9znrY0pbIqCWkSEqCYDWuybYZaiqPIJIlVKzE1Wl7Jry UBRtfbkiZ3Jn60bSghSYIU3H7EO8elg1lmJ8fRB5pjvGEuNGBEfPuSFHZN9N
cK0ywUt+md0RFPeeaWTASfkKnN42GwPvlCLAcGFOGcdUBs3a+HzCWflqSnsA CW9TR3bkAIhiC2GrUE16Lb9Q2AEpGhA9NW+avsUJqihC8A737FrQLKQO+Dnh
Sm2BtD09Z4xLAV2FvYNnSFBqs35AxLW6q5K+ed06Sa9YAKjzWGCXENhH2zNS NETRzFEZWepBVgXVrEO4CGZK/EWViTZ7wD0H6C8qDikFieKJyoWS9cpvNGYb
qBSTnaRdIwIF+LLligYaA/YgE+SNWphW06C+Q5lqbyVg1auWrA6x1DNfUs5B eJ0GIqlU/S08OpdhftOsBCWC4luksrTeOZtTSEWh4gnmNUnXgbWza9Opu7AU
lijv0Mg8SzNNNZZwTSOXUwJWV7BfSRKGKE7v06qP1HWGQwyJUG9JO9E+k10m 9rm6A8ZYdO9KYpk6rYFLi9ZAGRuEmFad+K28bFyUImdRZXhbYJHFYmz2QtbC
yiMNiJK/yK6ZcyOKkBooSqjeWZSfTC50anBBmdzWrTctYBFIv43PdXxuOJMA TqBoCe/pqRgts1/OOdXbc6qNGBjxg9RwDih+yzwCbsovEJFEO7iX8Ui8pmCY
D1IaYAXAVJr5BgBQY5enXBt1BPBUEqEVuLMHy1MJuZNynBHcPA5T67CnoJNY CSnwDJa3EQn2sF0o17ASGdwy6xrnNEbRYKw439Yp1SYHE/KaR4iUIzsTBxRz
ZM/p+jQkqPCYabUfDxGNWSIHcwWVq7V2JZExUc7RhLwuR8twAyXQzeSx9F0W AAYIEyoD+H/LrYa0KPBsanpR4C4ujWo4MhKUkDioHjhg7Z7l4P1dZqUKOdDc
6mZNMde59bqVUkzQPks7gryrUgN7rHEE0cCTAjCpG0NrldqZcpqEe+ICugxN qLZYJkYTKZrV4nxVKfywPMnQIBgNcpK36at+x47UzK3sjS+rnYTkLtmyI1+0
5YbGBh2/t2U5RwZo9IlChx/4fM8yT5Y+0yDoraY040zG91RGLgjeIjmRgnCa zMYNOzcNiBUlAxew0K4stFiWIKeIxrsEsOKtEV6St7e0/3KxU4+aTr+CNlDF
hiQw1gfPTpM+D2DK9CREB8ROheBlnHwxC/QMbADIVs1iQqtMYwNTtOIZq1XI NT6vJF2xeZcipYdPkyQpP/0GOvRbaNBvoj9frz1fqTuLnoy+RnP6evOOO88D
tyVJFs8vOqtjhWqSJBuqT7AD+2Hg1d4KkR0vT0gAFqK8LubN/rhduhp3jfhJ VmPd2q/h7zzdm5b2DsgR9Z98qL9PMlWY04RInrEALAn4QLIto1nD+7yrY8Em
nzTZD44FbA20g8TCCgcUobBgKTSjpz7AkdaVWgcYlF2i8btWnEiUanYmJ8tl B1jMY/6j8BbybqvgVkPF8J6AWmjisSOkVDsik3yjec4JfzRXjOk3NYskIczL
pEIC2S+elawINFGYkd+vJFKCkVRM4dQuZMixB8sylSJbiSAjgdRL5jSxijDN T8lmtMTV35hXLD2WRU0Tn08wQX4aUxMQZNcACl3ySZCWwKoyewPHLpiqMd2C
kVlym9TQsLcgzFMhJqW7tzADibR1Ncf6OilgWr+zuZqIzGNovlyH8PuQ+ESy AGZlWyRR87rzUjxgeavMQi9ZMVyHHVnp6xCml6LCIDVA9HX5gqoZ4WevziF9
uIVtcRO+N88+zWlSJjQL5RPuBoTFJ5uDAG4fU+4DAu8Tgye+YIJQUiWF4jGV AOkjqmpQ3vOuS8tS5aJhvgIoWLMaoHg62nuAlgeJ0lRBDNcUeHlNLsoCqisp
zZCP+KNHsJqFK5gw1jSNEytYB+WN7+1yqgqXCv8u8iBNeMhDhf4MYzOFtmJh cB+E0O5CRbMcrtDH+J0ibUT4rORIC60vAJHxNwhTeeMRBAls1T2o2NlCHY2O
DFoc1MeZ9odMS/H6Ypce4n03PezD1yncGEXGWweAk6ur61moYsRCq0BWb8AC lTLYntC29stVg6BOmB2fa/lcn5LgF6TuxfKWLbSsEyxfdV2eQkQNIkLoVwis
7JJThXSKLIGBKOWaGpSEfjdxCrBQ1aois6bRrfH3AyNOkSfYmg1cXSYQHL2N xeX93d0pf/xOodlK0N/2y0ahAaw0TM1T2jvVCALcGm3N4iFCCfYz32xYlyq1
sQ/pmndLdYmwwYcp0IRo7K7asVLBVSpbUcjFGgvZw/H1IlSpdLNBpH7Y0MV6 LiuwLiVTA3pq3Y0WmJMI1MjksU4soSZczSdKWydZKTMG2bNsKR53kSsqjfW7
uxvvNLA8vLR96qTrGRQQEiIJBKWdFAksnLHcPwivMGrPnahc9qdIFIavkxkJ wBnY0bD7kxGq3QQJ9RyeGEAp29AyRV/VIOL3SI4mrmZthU/MdfmexXdJ0s7W
HWxNqHj1pIeMD/cvJaQANrhG+8Fw02omGcWWquHeFUU5pTNM6WwMbcjCmBBR 6bsEw9VOk/Vhjl53SFwuiKtF2i3NjkSIpOauDpadyJbuPW/PUyYK6KcueBbJ
ZVNGkFjypsnEgBkmPw9pcTZdkUmJPU1VHVF8QHfGG4gZrLmrMIo1QhLzTe1y nx+E7AJgFp6tOIjlGgXKK+iiE8tYLEI1SUpAYvnz1ulaoVYqibRKFODWfehZ
Reu+eBzkDPFqSMKEjzEQnAhAyoBHCXHeyX1I81v1c9UroayfV4q24jvL4yj/ de0kDxtuTzAsNqJpSawK1dv1zJe4awCvu4KA++DZnNEQ20uYnaQwwhSW4wUl
h8fvXbBXpLNN0JHETyn7CLfYGYmPau8iUuYXrO52bfKloY4iYJsIagCZ0q0G d8gdfqTx2vbuCTZqv2/EiESqdmM5UiaByl67bu8s00Y/E3kZs9OFxEjg6YLV
lf4Q7gM2e2WFgxEFaGDYxKFLiIPL0TmzBMvsQDBTdr8EXXVkEeOIzCkLkORZ CVUMWXJowbLLmLYzgAz4Uc6YoMcS2Tgr85wDQpaavUa6NxZEkvtrBy0gA4qy
ua9kl3iUG3xiR3mob8k93ApV1r1qzDbswAZjwFMshcImul1hQoIRC6Oqp5AN mGB7rVTnXb1xmWqI0NHXX+5DstOQvsdUZw3V4nhap59dml6lTP4g1AabULbC
QZSrwbZRqnVwp5Yx7CCR66Cr2Yq9D/Y1GrvmBKaIbEBG20wTEkSnkmXq/lA2 5pPKgQGX98n2TvpZp/yTDgYEQkaFdEGGiZhBNl1va8Sqg3AFBGNP40iYzBnx
YMsD9v8kldWCA2uJqQi1Ixv6NODK3H+Cd64l+ZFdT6lLk13oCpd2QKU3FtPo xvduFvq+0rzaRACk6TqzKME9/dBMpi1Y9YUUe80f1rNCnUD99fEmPcT7Ljq3
WAK3Qm1ifUiTnrum86nK7TymW4Xdk7hnlQqX21ByHSpLgcxH4p1S9pVFEn1K D2Mnc2MQGXbF4E/Ozs4PQnkudhHEZ3X6K45dKgKhGECQwDiUaiUakyR5rCIJ
ouLek+5S4JNJiXzLaUTmMgg7RrSb+FV73MlmTnwsZo58LRcr3hT3lJ2EALhR 0FCVqrpmrQ41bOwlHU6BJ6iaC4mmvD+YeRMj36bEvmdqEGGeARRQg6jrvtiw
urvqeeWjtDLNrS9drlwlyPdEmC9TPxa0i7PB3EmH0z5vSMEFNogJmuDLAoaI BsdNKlZRl4stzqU7WZfTUIHVRpow/XZFA+vUbthFY+tj5rq8X7fTK40lfyQO
ECFntHzoVI3tVGqypRHqLGEndLvtqlgjGyqEtK0jSXKhIQmz133VVGWHGEwu KPUIJbCQYrm/F12h07WRoQDWwwgT+q8TigQMNjZUczvIQ7yH+2cSUuA0uEf3
TY6VZPJhW9DSRHKJh1tJR6SARi11BdDKFJsesXlPKBmOaTKThljxDOTpVMkR wbIdeyDp8JqSYVeWrBzTFsa0NYY2V0m3khKT3hKCSOzoUGlizAz0T0JZx4wX
BHaN2xLWU3Ut8H66OqYlEBQqoKOKb79tOaj0yv7clraiqVFkU2rkF20L+JO+ hFKiUWOVSOQg3DtDDjgNxNwWWMVZQYnZqvSZuuuuNxJYDQ5rVALN2xgJdngw
rsnbvgUhzmBV5zFJGO9vGWFCfzTp8T/KLi9gYOQDJHO9S2v1Lu5QjFsCFtlF Vf/PKFd7ua9c+0YtXUVLZ9bRlQKuWM9sO6hfweZvfNDYqXlTqZQkgkpJU8DF
1ii8DcvtsSTNzQRbBOIzKJqL8SF0IFo3Z7rV8FNb1v2OtaPWCyu7QXm9xss0 xkqEVIUXpjKzYO+ibZIx9aUUHbaNXg1eJveLXh8rBPzgnGtFhb0VxdNAtemI
LA7CmCSsWb/EflrqIsMwFullYdXApGwk1G44+Veu8e2M/CYRLQ+nFcOpmfRl TsAN7kZJZjGB2YE4TentinvVlYWLAzSnOEBqP4p9pTiCR9m8Fk3KQu1W7mGb
a8ggomxfKdVuD02MtKAYEC+k4QcWW2ioYUOkzyBTCYszNr2EoxaejEjF7+vu X1H3orLrMF0QdAFPsdA/i7Mt4iG6sj/FFLIhcHLRa4qmUh2nEBjDbiVy3epu
rltalh3f9TJ2MeVViCHxSxuzaZpqIJwiDA/Mz0mnD/PuNystiFoB+fhiCeyS 1qLxva5d5ZYkYIzIBtfoqnHyBdGsZJva/TQ9tNxD/3upSBxMWCuk81D6dGGs
L0vI4lqemEHpRLMcr2gkFsREoHAmG2icxKol02R3iwSC4avTXn/KLrdDIT2y DsbM7irscynJj3b02XUhutAdzlwPSkuBumWDxwm0ieVNTXquON3+pWmgMXe+
a9RJTwwlJQ9BDG9WI/IH/s7gyj6avQYUMxA2Fxn34gdOoBsCLDiKFGJ5cim7 cMiGdzFRbJRqS60px6OdgSHVuxr8AIOQXajO3h3l8VJOxlpFaEvGZnDqB6xD
mVRPW7KBATCJZxpyyxgim7ESevaCBQiUa2bH6tPQ7oiWItpTtFxZWyr0DEbu H6OvJ+pFZerX9oLELyaXqq4FBC4hU/4ROvUCnxX1SgCv2W6kVxpLDzF11eqV
2aH2P5Hk16VGDq0NMpCfw3aIH2K9wIBQVuIbBOnVRqEqCU8B3yICaRcfDLbV 7FgMOs5seIlDsOT0QNGh27vglmv0iOz6AQuyC5q8JxCcGSg7RvP9bhOGuFxi
bRyCz6gRQCKHPlTrW6KGH3M/r+mJ9JKAjDFmF4M8vmKmJRxVKtNy7xqZD7c0 gJZUYh1A/BedkxYZZCd91zQsBg3assLGVHT/ZSLiIvdIaSCibof1QA5JCCKQ
mOuSS6T6APOG1AamtUKpEZ9AQdyEJ3BJg9IZx+qlGJDppVQUscLJ5KKKeDsE v1cOSQhc5Y4c7P9gSYQRrK8yh6+1hd/AEL7aCn4l4++w/Vdpv0b+webhW9Xp
jYiGp8FpBI1OlCXVSbVaJLL1I9VJ9pBWkmRvTRvSvBNYTbbv/DfjBOL13fVV 7/JMGuehXB/mbdbrtoitlr44mD61zFZ8mOMG7Tq9k/q4YILNSBwI5+Nh+MTR
qif86V9Y5fy3PyFfWUPDzOODkH93fTPyuFgNhS1qU56Sgd9hMP77m9u33+uD VWcCTNcycSh9GMqonQM02Pmqw068J3Sehukq0/fYOAtZzK6IYzDeVH5NgJWa
jbC0tDWj78an1DCRm6Y59lQfEnwZ67EjyaTKU9SlOvoehmMUzCJogvfPYof0 NCEDZ8gFWQIFQiNt0DrshmN6LUOZAllTU7RIEdMaDTbHTQMYIuPwozfdmFuk
FA61ZztbIEnTJDvjQ4KpVUj6myZt8qU6tqCTeCtZrrg15j+4zQNfKO3Viq3E YFFmMV0fTlFYSUn+atPjfz0wOozKkDywAmZWXXjVRlDsgw+Hz6bm2FQKNfot
w92lv66KZB9bqjQ4KC/nrhUsOMxE2FVQc2hzFDyIRczYukUhj3ZEvfvRqnuu 3djdZNPazUMW0mvHivpdeuBSW+1rS/tLw1HdbJSn5OdO5g6ycln4nxSh9hCl
mPU+iWkTTJ6NJjRCTdWVi0nRAPZd1kYSFNee9ZsjMF1tUgA52JscxM+TM/YW VI9Mt82OLDWSPqKMud7cqZJJCVfyrD7xL3xVNwfMNlLWU8NsRXlKlmDMEjyI
ocCNXPF+3FgCOlYFd2P0ODvtrhBix6MwfhiZYkKderRWqWKPSSzYT9kb8+MV iKdruulUoRYptDcV0Eeoid2y8EllDY33rp6Tysmk2HYcjlLYG0D8v5TtVTtz
A/WgEQespLbWMO5zv6SLWydVaNJ6b5fQ3PrUI9UeombVbhixYmJixiGqVh4q 7GD1eOxjAUqdDLOwNAKUyFQlIYlQPqRhXiZKWQU7XWhvzQngii8WjC3VK4GP
GxnAXHDicdYV86zTlgVJRWuxp8/BI7Xat05GcZwsUtE/4sJBwHoQdgrL/gQX 3Mue7ZUxteRQqz8SDWJWPvdWWwoqcaY5DdM+TgMLKOu/Ok2VpUrPus+ke6YT
NpZlsn1DntaqpBXKx82irk37I+xfPRG+IEkqfwt4pHcbaV+D2E18HUGgYfvw Wpm9JKfkIbDhdDFIxZBME+hy7vhGwZ3tMZubjFNfPSPQgM26v3AhdglmMjdD
si5coKFf8DCE051axNxKTUKKqJou7I6NEB28AaNLNNDE0kfdmxI0okh1/YgV 8TQ5R+XgKvFMJdN3av07MugSCdAvvlyrLCwE99WODlM4u+swF87l6n16K3d5
Ru7DyLIxLvw11Dd93NFkIXxHM/X2ZGGS0ca2vZ6apARytwGxBPjsNpjGFmQz muQCknCXuYYOLdMTUx+Zt5YuRJQXbiCUePkGcfaqopCUxKfg4qIT0hME0NdG
VyNiB5eUniJhmsVZMDozsAN3N1J160MaETFWKabIrdbs94HgpqnreZQwO4k+ BwIE5vRpk9ChD6mjivK9z/hqrRTIzCLSIqLnea+iVrDoIcmidIfk3qX1LAxK
zBr4YMjUJ5MfQu9KWnUBdtthNX4cP6o+cxLbZoCOu2WhkU76KaV6srRagukb 2YmoPlXqmMOnYWOt2kuTZscRxGEvui0ZhN3nWh0Tg196LrV9ditHx0X0uH2X
BwiSYHY9N8xL47bJwMLMDKMutwQFzEKS8kgjjGb0jyIa636CVXaEVJDyU6Hg EX3hbngaOEYvspI+geqsL0JvaNgn4OGbYhtOqISKy45TTZrv6z8Mk/mXV+dn
gkR/E/yGZnQx88LuyloyvYF+sO5++4PsfJG9l5qhVJNCgTAVfUU7T3QXgEVR qbT3H39kv+FP/2Eau4SAWVILTH7+AS/Uesh5aEmYi2BzZu9fzy/2B+YY2xbQ
uHpZS/cEjxVwx0V7YlPm7lMVe1CnE+OJSbfsxLBsELujtpFoYmle+6tkD0vT VJ0MV6zwr3gX//vj5ZvXum4l6VQaAVDS5FVhbi+zVbXtUvLdlnqNREjKHsIu
gjSy1KfG0Z/xXaRXxU1WTDMP7X1Q95Jh8zggK9J6LlkfXdck23uQ+e3T05L5 Mcax1jGq0F4a9/g7/jOd0BbRcLT+7NmKMIi1mw89j/2agbxSZTpqmDqfG2iz
SbCJdVmP1wDu85Q3xmIQ94ZN2WdLu8+DXGhqiW6tdFa4Hrs1Oa00A2e1k4DS VQcbHTlSrANJivS84Bh2fsNx7oDexkmotg670U4FCz67W5Schn5T/AhzYXE4
FoYnWQX301wu+DG8M43ogsfHuQSvP83ahv1emnolWNImL7ndbJdu3dXswitc PL3R3VnD81ETFgueD9tJKP7/7OaXbyZ1MT+/n7SddO/ujgDM8eBo9Fts5u/e
oUmHkt9o0YMmbjIJGxs/tcE5TfVgjp4o2ScVcIRaOmovYudQJQWZkN0Om+Ox SU8sX9jDZ0TyK8v/7r6NytylFlQ4LQOHFwgRMB68Qjh7IdEj9p/iRDmNbjCJ
BASVXjOxUZhNx4UAJbU3qFmH9lutRehxiA/D57l3UIWqpxzMyNnPKd292WU/ VQMDqTNfsGC5F8td8JCcf6XP0iKrgncp98Id5qWVypJv1BkER6djkwCTNzZD
hZNo9fgL6kpAJrbO7JBNzMY1rRjCPbADwMGgv8gu9GhEA5Nq7rmK6TgpWjZs mlAzw+gchAb53Nnr4bgr4HsRnDORxv7uvKckAjztX/dRTCyFpsnxReq0gogp
dgrepp/YtjCdTK6kMaLvmTiOamYrlpSF9Jfaxd03RoxLl2P0MEs400yf0NlH z3h0ru/+Wq/620HOUEhPpCJGpJNoY8u7CKPjPL+/EV858N+qDVGwqjVEN7Gg
IKio7FDz6nurYyYVQe7gYKNLG3abw56kiVmTCjtuDA/LHGp6qVfGfG4HMpYm ZIdwptS8RZ4EMkBoGVbLYn1sd4xSSoilWvhnYimF2h3oiOzY2aRChRhF5BhY
x8lK6gGJPYFnGvVjWO2HZyJp2Obj/GhjizJx1UwbElQ+Zyc9VT1GsD85EIlg H6LMHackfZhnE2K7YwLaZBgMaENpU1+bZ2p2mC9xJzUuJdSkd4tf4KSBja9j
C6c9Vtppu2f51O4GXQraDDsAfQ1aqaT3sDov0etktWHfNbmc7Lb3bbXJE+xe TKh4pGlWzn1IW77hkWivA2LAZ4WqhDS/NL3cbCsBxXgDVhfGakGwjrK3OQtg
9pR6YiVlXoHWx96Sdkkfz0lpvaHJWLxla20cVXGbDGxMcbWIE1t0IvCxHi0J qSMbHYWV+7CyzONJvhMaU3WcQWELc0M1rd3OxqQSGQ8TdDA2Ff42KyQh8Dyb
vWwVps46Ke/X0DqJFDOALXfe4HGytU/rlMOUJqXEqeUrHQ1YWiSDiv43KTO7 FchYIzHJVIk4Vy5NgwiuDyIVhHJEgYjSK2mXdACIATJWl8fIxJecQAbjxml+
TY1h7OPVGpr2C2sKYCPISeVaA/y6rmX4RhNZbZSj2EkPxRDOJ5N/GFQX+Ry5 dlDo9IJVmGXywVBhHY3ehVnatOs5MqF2k4cCZ0IbRZdpi24TzaUpB5nul0Me
tdawRq053PPqLV3KVH2yGaR7Nh5sD47dc18MwNfGk/alsjOhn6Gn5PXd3U22 UvSeOS2cd+OK9JBIAro0IsutXyf9CoRZQjROcogrCzntPdO4Woi915+xYSOe
AezT9DHUCxFuNKfEOnIKvoe/BzdKKNMt2AAO32T6gtDyFtQzMp1oUymX+0kr yg38FJh8KNmaOKJf5XzD+TjR8rnb5KUUBnriCTamOdzOSbOpeS99HwljocmT
AvttA2HX+hA8cEj9g3x5vSftgbMLaNy+eCtcUuzHjqWkdzyQzfXVWWDhb7UH +nYipz3t47KvBaPPSxnfFL+S5xM9s5MqPr1OZK/ZInoUqzXSTme9KQ5vr2N+
7A2XTMtgO7fOyHg9Q3IC6i5ONlSQ4HKrrpQGTNhALEacWNOsbxgRa1pjalyB gm3WOv4tgwiaTaalpcUwhI0EhsLJIs7JgM4snD+A5GcMn9seypWTcVIwoBXb
mMxLrbWfOtdJwRU3hozhOF8asfUAtLz0JuwUhw4Buhm/CHsCqS86kMgxeNDQ pIZ3igY36em/sskh3IvNtRqvgefPUskh1vM53mPzLsnefN7fhbHaaOGaBkmS
pKqvvVaVl6az/DjS0RdgDQhk3Be/HJ28evRQRMLFeI5BTiWE+q+TX1QIG4dh wOMkJCtR4F09mODWZJRDET4TV9K/M63og/FHWoID2E32D3rjypqxJw+lU+dy
zJOzhIy3cn6TJ9sk9l6kNuPYev/TB3R5poPK0APbX89/8bVMIn786hk18t6V u13P/LIteURg7ueaq4Zh76DdvVNmxBQunkzRE1iJ1Fu7rekwu1wURlHKkZ/j
96DruO05Q22TfVeZzrKLInuv6UqLay8uZ9kPtxfEnPiGdCIlcCF/b8M5ziWj OLtcSCUvFEX6p/ewBcSXTjLxJBNPRc3FZ1J6vcZjOB+kZSw9rfmh/zz7v0Vo
ZDxVgZC/kwNVWrHUT9q53MWOpFi9unjOH6iQAswf9OD/fw0Ps4Sh2Ufbn2px XMm50YwHTuT4kTnpSNgJXPe/oCzE38Th3Q2y0INhMTRG8xp+BE6E8X9qjvXk
7ckRlrtT3hoOGTHh+6mTRX8YHSzCi5lvp66NwRGiwb76qE/zrz1XJLXfEAXC ZgWVqq65i/EwmZ5VHLcO5qb/wnvGoPFMZtu6sbftoNi6YFNQssVcj5l1s23D
9bArg7BZmGBMPp7ZUoBXAeiBLiObv+Oz0HXomC3MMXj13aZrPD4O1M9+mLYd 7tPQk9gZjOlAn1DqoycoZLJfy6Xd4a+YgUeHd+uhozNahLQAdazExmxbmR2n
HL5Y28l5dqEdJmxKijb4E53OM+Fzar8awhIzFucfHUIEHSPTUDGa8t7zdcMT e/rVMVW9NOhoPzdEEmvawyw3jfHFcwn7CgBihO2WZwHCckbT14PhBPLEFwc6
FjSGXWyg1PM0A06pC3nVuOEyvno2mbzFXIhafXA4H0YTOVhpj9A73jd21Xh8 U6b82d+Z6u58hB4xFzYHXdid8tZjQDdMCdymN2mmJ3V6AUDjV6oG322wSiDb
8MF+lYxkRx2lJ0XVyaTPnc5B7Ew7ONsJyMC73msPVV8Ee5gGBT6gr9lbZGEv 2W0YnUkmJyNT3ZmfZAnuRgYDdnox4lrve0uadLm/mEHtDaegxFrWzsVV1W8T
xifaD43ZzbsdRrs9Vtysdg824r5Mh91j43U4NC0a0M0hbnCXRdztC5768Yv+ jA3Rbqj9QTqhJ70IZ6E4p+sWOu+RhvulRVtC7ARVTAbWHJ+AycmAFtVTvl9i
HBlSi/T96UYfJcJTZoNWPtk00e1B9pyVLhxC6gdJja4PjyQ+ubk8S2z058+c UzElDeymw4vSQFT3f5GS9ss01stzRlp71SRaswEXvVz6LIRUYJZlKW+otAqi
awroQaaE/S8YIX4Do8ET9/f1LHvZYO2ALHz92jU1Rrk3SBmMm4V2KIbd78GJ 4/pkPdGiKMPRaPSPveK0nAxwM61+DgYsObvQKbsUOLtKRGDw/nCtGyDuDglj
uR7c9Mo0P5rKbgBxhbO464VbO6YA17bovJHpyW3LhtWHS9xbzrJrlze1r1dt Ab41fngsV6wmYDSMBr68urowK3h+aj+WeibsjRqVQEjGGaLOA965UaKZTtIE
vH5lK/5MBaIKpvDqFWMGhOT5HYn9DNBJ0c+/tTWbhJn/v6ngGi130DHGjAjH /3Bg9AVhZDkIaKA9Ua1SYvdFRQIWbgJ818oijLCfCAT28noH4QOCF7/RyfMZ
ga5NczAV3ndkanNZ1l2xglVZXLrcNJB7vWPIflWaLbLaNd5WerBW2/AG05TZ jDMvN/KCNwN0v3f57A28AAgWPXNDXuo6dzh4frYfkPsbnfc9JWOoQTyWpnTb
t2ZTLW2zlrE+ZN9CA9JZTOz7nufaLusGQUdUhHvuII7XzBb5hhfci3GAT3z1 Wk/D7nh/H7cUSpSwzUWbi0ZAUWK5a0frDrrhQMGYS5DGfYpePdd+zq4V7hT0
o/nL/5TzawPrs92HWfb+l5e441sq+U15ZNb6bV2vSxslcGsas8l+Y+WAPU8T cWPIMraTmRWbCB6Zl07DXFAYB6M98ofQd0qHuALyHHoZqqN0jnSutqhlwDjb
fQeuRI+VsMdofSvbUZYnt6p6359DuGNuVJf1mvzqBrlG9h3yds757Q7c/5XY DiT5jTk9fn3MWNk7P757tDOML8idNounI78BMkGw5PzUlx+XcLDzOZtwNDO0
L4lACUhEOrPBdYBhST3Y1vDd4x/rgdxIscZydc2fTfbdX/53U7LBsugtZXTb JuAobDiPWcc1d0EuYrp814TH+yW+H6fDVPH04Zc/XMODrZSjfobpyeShfpIq
t2BxUOaO519n2XPb3EPK2Xel5c+V8ABTk27CIGaDOeiPk4TO6FepMUneidQN /vPRAwrzvc+vkR7gtqcM55V5VdjWcdrOvNfsqKF/fjY9gcK9uzymZ4svSSdz
9937PdSJ9Xzr2tfdkuZEK9zjqat6gxdUutSLN1zL6LdCZvoDKDdSsZFCl1Ix A+Sqr134msWMwTieLAWy2MjBci2o67/0iFYbp1djdfX4aXRI5uNH/TwaTwN2
oqIorKLFtLDbd2YXvlRNXOeQUmVmMMTQWEEdvLN+R9BjPTz8tEoZjgB/AwoD x3rD4o2MWsTzvb7ZOcx7tQuQw3FrJpm/8ox1+C6LGk3vOHVvCGsw1f9Lz1hL
1Vdtt6WwIZ3s+45C1nXL4y/stgZbAvl5oDixTliYLRA8Ibrnl2rNwx95Ghp0 dyIEnHA99A4Roec26FQdz69rKFEW6OF2K6NCw08FleFkxdxug1+4WrVVjX8e
tBtR+NsNrg1844rn9PX688Z8IM24sfx1BRjW988veVifGZQp04tvLOItSCUM 9NSA85NN0zuHunRHoyNzrCOJHGKNyviFg10HAh5VkTXfTjBcHMjgkwzAfoQ1
1d07KbpTFBZcEOLC47VneeHK5bmTBcKFN8aWFC1g4l4m27bZu3qNgZkSvnIU ykibX9fyvv5RU+rEJk7c69HiHoLVvbyo/GAnjx6MRm9ADX1fF4iO+pFLPjTh
OFkvzzQQK25YJYd4KzaTczVvnl8TfpAWN0jHbm2BZB3f3txdPr25+uGWnLRG tpA+3zg02/g9hTttVVnKDQ4h7FT/R6MuVzvSs2Xdxy7gefiy9zp42xXg7uZd
8g+Vlq4daOLVz2niuWbh0mcP9or539a0tWQlovSK/YNbmex1/SNWz5v5VfZb AYDoe24c0r5nw88+3VZ2M2k3XO5yW3DCyd/pGv8ufRIqntMJX5ERMWgXk1NR
00He1WeNMwr3O7I8wCkkc8lSMxPw8GM7gqssdrzfHIm1wdlOksFRve7BeWJp +Ty2poPRfvymO1qPZCb9vtuVJk948L43/y3dPe1lc1I59+FMdrdIOh1x9ysN
uRXq3R4juR5EiJgjNHZvDZI0zRCRSkK3Veig0l+iEI5LDngtv6mQttDlAAo5 excn+wn//vxHeDTprAHfJN+YMtR8/PjxR+gOHrq+Lj/RTJ9X4ADcmF586asS
CEvye9z38Vx/N80W/zpdYWF2+onbB5+PKtvxgNIbN/6pFzlVJeOnU6LLvn8L y11bZCvWyw06UMuI/xqInHvTW1/Y6idbuBX839w7vfeZX3qmIedu3tZWCI43
r8K052I8G7Ck5zUMbMKfWiNwvDDVsXQHfH6xabo9/l8XR3z6Dewne3O01do0 zyoWRE7wRC63nvusKuty0fTuOnMFv1KHyKWkvXihvx+DizWvMNeQS68Kimfy
+HgJPwPSRfDEN6c/PzZ/AGD46r0hFcu+6/jTODIq4ezKcRsP6bjMavwja/Ps gyt5+oRlidMCNtRwIATrHdAhhkXPbXVrC/PCbt1cnj3Jy3a+yHm0TG44WVUQ
Dto9itNxHhclCFcNW0TGb/D5P0jtO80o8tQ6J9z2cdHSGm5CTL9MMT2Fc7J2 UrkhUHiR2zWS7qW+P6+Bq10VbrNVbn6wq2LmKl6Pq38wP0B0eSCMDvQ1PxFw
yQa1EC2bCNX9ZxOLUFuG1UiViGnCUVmZT9Q0nReXGoe2ocigdddkPgfpDltn UlYIX3pqUe68AgdfMr116eFnbD16OGNc+Mn+93/lk3MLHXbtB1nr/eMTve8H
ejaZj4Vj1VLURDjz6SS5jedOUjqfDkiLLf6nkazgoqlMTdOsClbAX4DRCXAB 6stpvuWnHPD7D2W5zF2Pb5e2sivzo5NPGOk9714BMtETaBUXDvVSmrGOx+OL
cfL77Pddc3+c/B8blsATJ1AAAA== 8qY7CnfFDK/My+VWF7tA0mReuTzu6M0GecwLsQy94XkOt4vsbIW74HDzIE/X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--> -->
</rfc> </rfc>
 End of changes. 107 change blocks. 
413 lines changed or deleted 517 lines changed or added

This html diff was produced by rfcdiff 1.48.